Eleven cats were recruited in this study. Cats with blood urea nitrogen (BUN) and creatinine within reference intervals were classified as control groups. While CKD groups were classified into stage 2 to 4 based on the creatinine concentrations in the blood. There were two cats with IRIS stage 2, four cats with IRIS stage 3, and 1 cat with IRIS stage 4. The clinical characteristics were summarized in Table 3-1. BUN and creatinine were significantly higher in CKD groups compared to the control group whereas age and body weight had no significant difference between the two groups.
3.2 Concentrations of indole and IS in CKD and control groups In order to investigate the relationship of CKD progression and indole metabolism, concentrations of IS and indole in the plasma and in the feces of cats were detected by HPLC, respectively. As shown in Table 3-1, IS concentration was significantly higher in CKD cats compared to control groups, while indole concentration was higher in CKD groups but there was no significant increase. Additionally, the relationship among IS, indole, and clinical parameters were tested by Spearman correlation. The results indicated that there was a significant correlation between indole and IS, but the correlation between other variables were not statistically significant (Table 3-2).
According to the IRIS guideline, cats with CKD could be classified into four stages based on their blood creatinine concentration. The concentration of IS had been reported to be gradually increased as the severity of kidney disease increased (Cheng et al., 2015), and indole is the precursor of IS in the intestine, therefore, we arranged the concentration of IS and indole in cats based on IRIS classification. These results were shown in Figure 3-1. Cats in the IRIS stage 4 had the significantly higher concentrations of IS compared to other groups. However, there was no significant differences in IS between IRIS stage 3 and IRIS stage 2. The IRIS stage 2 had a significantly higher concentration of IS when compared to control groups. On the other hand, the concentration of indole in the feces was gradually higher as CKD stage increased, but there was no statistical difference between each group.
Table 3-1. Clinical parameters among controls and CKD cats.
Characteristics Control CKD p value
Age (years) 2.5 (2.3) n=4 12 (8.0) n=7 0.08 BW (kg) 5.17 (1.3) n=4 4.8 (0.6) n=7 0.32 BUN (mg/dl) 24.85 (0.9) n=2 67.7 (49.2) n=7 0.02*
Creatinine (mg/dl) 1.65 (0.1) n=2 3.6 (1.9) n=7 0.02*
IS (mg/L) 1.41 (1.0) n=4 11.27 (17.7) n=7 0.00**
Indole (mg/L) 23.4 (21.8) n=4 32.4 (60.5) n=7 0.82 Variables were tested by the Mann-Whitney U test and were represented as the
median (interquartile range). *p < 0.05; **p < 0.01. Abbreviations: BW, body weight;
BUN, Blood Urea Nitrogen; IS, indoxyl sulfate.
Table 3-2. The correlation between indole metabolite and clinical parameters in cats.
Variable Spearman correlation p value
Indole: IS 0.622 0.035*
Indole: Creatinine 0.203 0.528
Indole: BUN -0.175 0.588
IS: Creatinine 0.524 0.084
IS: BUN 0.252 0.43
*p < 0.05. Abbreviations: IS, indoxyl sulfate; BUN, Blood Urea Nitrogen.
(A) (B)
Figure 3-1. Concentration of (A) indoxyl sulfate and (B) indole in cats with chronic kidney disease and controls. Solid horizontal bars represent medium values and boxes represent the 25th-75th percentiles. Whiskers represent the 5th-95th percentiles, and open dots represent the number below Q1-1.5 IQR or above Q3 + 1.5 IQR. Statistical analysis was performed using Mann-Whitney U test to examine the differences between each group. The different letters (a, b, c) indicate that a significant difference between groups (p < 0.05).
Control stage 2 stage 3 stage 4
0102030405060
IRIS stage
Indoxyl sulfate (mg/L)
a b b c
Control stage 2 stage 3 stage 4
20406080100
IRIS stage
Indole in the feces (mg/L)
Control stage 2 stage 3 stage 4
0102030405060
IRIS stage
Indoxyl sulfate (mg/L)
a b b c
Control stage 2 stage 3 stage 4
20406080100
IRIS stage
Indole in the feces (mg/L)
3.3 Diversity and gut microbial composition in cats with CKD
Many studies have found that the gut microbiota was altered in the human patients and animal models with CKD (Jiang et al., 2017; Yoshifuji et al., 2016). As a result, we performed the metagenomics analysis to investigate the gut microbiota in cats with CKD. The relative richness of gut microbiota at different classification level was shown in Table 3-3. Compared to the controls, the relative richness of gut microbiota in cats with CKD was decreased. More specifically, because the diversity is concerned with both richness and evenness, the diversity of gut microbiota was assessed by PD (phylogenetic diversity) whole tree and Shannon diversity. The results showed that the diversity of gut microbiota in cats with CKD was significantly decreased compared to the controls (Fig. 3-2).
Principle coordinate analysis (PCoA) based on the UniFrac metric revealed a separation trend of control and CKD groups. However, cats from the same living conditions did not show a cluster in the PCoA analysis, which suggested that the gut microbial composition in control cats and cats with CKD was different and the effect of disease to the gut microbiota was more dominant than the effect of environmental factors (Fig. 3-3).
According to the sequencing data, the gut microbial composition of each sample at different taxonomic levels was generated. In general, Bacteroides was the most abundant phylum in both control and CKD cats, representing 54.20% and 43.20% of total valid reads respectively. Firmicutes was the second most abundant phylum in both groups, with average relative abundance of 36.5% in the control group and 29.20% in CKD cats. However, the average abundance of Proteobacteria was higher in CKD
groups compared to the controls, accounting for 21.7% of total fecal microbiota in CKD groups and 5.20% in the controls. The other dominant phyla were Fusobacteria, Actinobacteria, Cyanobacteria and others (Fig. 3-4). Based on the average relative abundance, 8 oders were dominant (> 1%) at the order level. Bacteroides, Firmicutes and Fusobacteria were more abundant in the control group, while Proteobacteria and Actinobacteria were enriched in CKD individuals (Fig. 3-5).
Table 3-3. The relative richness of gut microbiota in CKD and control cats in different classification level.
Sample_ID# Phylum Class Order Family Genus
CKD1 6 12 17 32 43
Con1 10 19 24 48 58
CKD2 7 15 19 39 47
Con2 7 17 25 47 65
CKD3a 6 14 20 39 60
CKD3b 6 14 18 35 46
Con 3 5 13 18 43 50
CKD4a 6 13 17 38 46
CKD4b 8 14 17 37 51
Con 4 6 13 20 35 40
CKD5a 11 23 26 57 66
CKD5b 8 17 20 42 64
CKD6 8 16 23 42 57
CKD7 10 20 31 64 98
Total 104 220 295 598 791
Each control and the corresponding CKD cat were from the same family. The different letter (a, b) indicated samples collected from the same cat at different times.
Abbreviation: Con, control.
(A)
(B)
Figure 3-2. Alpha diversity metrics of (A) PD whole tree and (B) Shannon diversity from cats with chronic kidney disease and controls. Statistical analysis was performed using Mann-Whitney U test (p <0.05). Abbreviation: PD, phylogenetic diversity.
Statistic comparisons by kruskal.test & wilcox.test
●
Statistic comparisons by kruskal.test & wilcox.test
p = p =
Statistic comparisons by kruskal.test & wilcox.test
●
Statistic comparisons by kruskal.test & wilcox.test
p = p =
Control
CKD CKD Control
Control Control
(A)
(B)
Figure 3-3. Principle coordinate analysis of gut microbiota from (A) cats with chronic kidney disease and controls, and (B) four different living conditions. The first two axes of the PCoA are represented with principle coordinate axis 1 (16%
variability) and principle coordinate axis 2 (11.1% variability). Abbreviation: PCoA, Principle coordinate analysis.
(A)
(B)
Figure 3-4. Relative abundance (A) and average relative abundance (B) of the gut microbiota at the phylum level. Microbial compositions in healthy cats (n =4, H1-H4) and cats with chronic kidney disease (n= 7, CKD1-CKD7) were based on 16S rRNA sequencing. The different letter (a, b) indicated samples collected from the same cat at different times. Others represented unclassified bacteria.
0%
CKD1 H1 CKD2 H2 CKD3a CKD3b H3 CKD4a CKD4b H4 CKD5a CKD5b CKD6 CKD7
Relative abundance (%)
(A)
(B)
Figure 3-5. Relative abundance (A) and average relative abundance (B) of the gut microbiota at the order level. (A) Microbial compositions in healthy cats (n =4, H1-H4) and cats with chronic kidney disease (n= 7, CKD1-CKD7) were based on 16S rRNA sequencing. The different letter (a, b) indicated samples collected from the same cat at different times. Others represented unclassified bacteria.
0%
CKD1 H1 CKD2 H2 CKD3a CKD3b H3 CKD4a CKD4b H4 CKD5a CKD5b CKD6 CKD7
Relative abundance (%)
3.4 Difference of operational taxonomic units between CKD and control groups
To clarify the difference of gut microbial composition in the control and CKD cats, the average relative abundance of observed OTUs was performed by Mann-Whitney U test. The heat map showed significantly different OTUs (p <0.05) between CKD and control groups (Fig. 3-6). In total, there were significant differences in the abundance of 222 OTUs between CKD and control groups. Specifically, 14 OTUs were significantly enriched and 208 OTUs were significantly depleted in CKD groups compared to the controls. OTUs that were significantly enriched or depleted in CKD cats were further summarized in Table 3-4. OTUs that were significantly enriched in CKD groups were classified as Enterobacteriaceae, Ruminococcaceae, and Erysipelotrichaceae. OTUs that were significantly depleted in CKD groups were classified as Blautia, Coprococcus, Roseburia, Faecalibacterium etc.
The results of linear discriminant analysis (LDA) with effective size measurements (LEfSe) at genus and species level exhibited many biomarkers for cats with CKD and control cats (Fig 3-7). Enterococcus was enriched in CKD groups while Faecalibacterium, Adlercreutzia, Sutterella, Odoribacter, and Turicibacter were enriched in the control group (LDA > 2.0, p < 0.05). At species level, Eubacterium dolichum was enriched in CKD groups while Prevotella copri, Clostridium perfringens, and F. prausnitzii were significantly enriched in the controls.
In order to look for the gut microbiota that could be a potential biomarker in cats with CKD, the result of NGS data was analyzed by comparing bacterial abundance in cats from the same living conditions (family). Venn Diagram showed bacterial groups
which were either enriched or depleted by at least twice as much in CKD cats compared to the controls (Fig 3-8). Enterobacteriaceae was the only bacterial group that was shared among all four family domains and was enriched with a greater than 2-fold-change in CKD groups. In contrast, Bacteroides, Sutterella, and F. prausnitzii were bacterial groups that were shared among four family domains and the amounts of these bacteria were two times depleted in CKD groups. Table 3-5 summarized bacterial groups that were either two times enriched or two times depleted in CKD groups and were shared among three or four family domains.
The microbial community analyzed by next generation sequencing demonstrated a relative abundance of gut microbiota. To confirm the results of NGS and assess the changes in bacterial quantity, absolute quantification of gut microbiota was performed by qPCR (Fig 3-9). Bacterial copy numbers were converted into logarithmic values before analysis. Quantities of universal bacteria, total anaerobic bacteria, F. prausnitzii, Lactobacillus group, and Bifidobacteirum group were decreased in CKD groups compared to the controls, while E. coli was increased in CKD groups. However, only Bifidobacterium group had a statistical significance between the two groups (p = 0.047).
In CKD cats, the abundance of universal bacteria and F. prausnitzii was decreased, consistent with the sequencing results.
Since the concentration of indole was higher in CKD cats compared to the controls, we further determined the abundance of indole-producing bacteria in two groups based on the sequencing data. To date, at least 85 bacteria had been reported to have ability to produce indole (Lee and Lee, 2010), and two specific bacteria; E. coli and Propionibacterium acnes were identified in this study. Compared to controls, the
abundance of E. coli and Propionibacterium acnes were more prevalent in CKD groups (Fig 3-10).
doi:10.6342/NTU201802223
38
Figure 3-6. Heat map of the relative abundance of bacteria from operational taxonomic unit (OTU) in CKD and control individuals. The columns corresponded to different cat individuals, and rows corresponded to bacterial genera. Red color indicated increased abundance relative to the mean for OTU. The different letter (a, b) indicated samples collected from the same cat at different times. The statistical analysis was performed using Mann-Whitney U test (p < 0.05). Abbreviation: Ctl, control.
296442
Table 3-4. Bacterial groups that were either enriched or depleted in CKD cats compared to the controls.
This table was expanded upon the information given from the heat map.
Variance Phylum Class Order Family Genus Species
Proteobacteria Gammaproteobacteria Enterobacteriales Enterobacteriaceae Firmicutes Clostridia Clostridiales Ruminococcaceae
Erysipelotrichi Erysipelotrichales Erysipelotrichaceae
Proteobacteria Betaproteobacteria Burkholderiales Alcaligenaceae Sutterella Firmicutes Clostridia Clostridiales Peptostreptococcaceae
Lachnospiraceae Blautia Coprococcus Roseburia Veillonellaceae Megamonas
Dialister Ruminococcaceae Oscillospira
Ruminococcus
Faecalibacterium Faecalibacterium prausnitzii Clostridiaceae Clostridium Clostridium perfringens
Clostridium maritimum Bacteroidetes Bacteroidia Bacteroidales Rikenellaceae Butyricimonas
Prevotellaceae Prevotella Prevotella copri Bacteroidaceae Bacteroides Bacteroides ovatus
Bacteroides caccae Bacteroides uniformis Actinobacteria Coriobacteriia Coriobacteriales Coriobacteriaceae Adlercreutzia
Collinsella Collinsella aerofaciens Enrichment
Depletion
(A)
(B)
Figure 3-7. Indicator microbial groups within two groups with logarithmic linear discriminant analysis score higher than 2 determined by effect size (LefSe) at the (A) genus level and (B) species level. Histogram of the LDA scores computed for differentially abundant bacterial taxa between controls (Green) and cats with CKD (Red). Abbreviation: LDA, linear discriminant analysis.
CKD Control
CKD Control
LDA
CKD Control
CKD Control
LDA
(A)
(B)
Figure 3-8. Venn diagram of shared and unique OTUs in the gut microbiota of cats. Each family had one CKD cat and one corresponding control that were in the same living conditions. Bacterial groups with fold-change > 2 which were (A) more enriched in CKD groups or (B) more depleted in CKD groups compared to the controls.
26
Table 3-5. Bacterial groups shared among three and four family domains in CKD cats compared to the controls.
This table was expanded upon the information given from the Venn diagram.
Variance Phylum Class Order Family Genus Species
Proteobacteria Gammaproteobacteria Enterobacteriales Enterobacteriaceae
Firmicutes Clostridia Clostridiales Lachnospiraceae Blautia Blautia producta Clostridiaceae Clostridium Clostridium difficile Ruminococcaceae Ruminococcus
Oscillospira Erysipelotrichi Erysipelotrichales Erysipelotrichaceae Roseburia
Eubacterium Eubacterium dolichum
Bacteroidetes Bacteroides
Proteobacteria Betaproteobacteria Burkholderiales Alcaligenaceae Sutterella Firmicutes Clostridia Clostridiales Clostridiaceae Clostridium
Ruminococcaceae Ruminococcus
Faecalibacterium Faecalibacterium prausnitzii Bacteroidetes Bacteroidia Bacteroidales Prevotellaceae Prevotella Prevotella copri
Porphyromonadaceae Parabacteroides Parabacteroides distasonis Bacteroidaceae Bacteroides Bacteroides caccae
Bacteroides uniformis Enrichment
Depletion
Figure 3-9. Absolute quantification of bacterial groups by qPCR expressed as log10
bacteria per gram of stool. Back and gray bars represented control and CKD cats respectively. Values represent mean ± SD. Statistical analysis was performed using Mann Whitney U test to evaluate the difference between two groups. * p < 0.05.
Universal
Total anaerobic bacteria
Faecalibacterium prausnitzii
Lactobacillus group
Bifidobacterium group
E.coli
0 2 4 6 8 10
Log 10 number of gene copies per gram of stool
Control CKD
__ *
Figure 3-10. Relative abundance of indole-producing bacteria. Back and gray bars represented control and CKD cats respectively. The results were analyzed based on sequencing data. Values represent mean ± SD.
Escherichia coli
Propionibacterium acnes 0.000
0.005 0.010 0.015
Control CKD
Relative abundance (%)
3.5 In vitro test the integrity of Caco-2 cells by indole treatment
The disintegration of the intestinal barrier had been observed in rats with CKD, and that may cause the permeation of pathogen into the body as well as trigger the immune response of the host (Andersen et al., 2016). Indole is the precursor of IS in the intestine, and since the concentrations of indole and IS were higher in CKD cats compared to the control group, we tested the integrity of intestinal epithelium by incubating Caco-2 cells with various concentrations of indole.
To determine a nontoxic indole concentration to the cells, the cell cytotoxicity of indole to Caco-2 cells was assessed by MTT assay. As shown in Figure 3-11, the cell viability was not significantly different between control group and cells with 0 – 10-3 M of indole treatment, which suggested that cells could successfully survive under these treatments. While the cell viability of Caco-2 with 10-2 M treatment was significantly decreased (p < 0.05). Based on the result of MTT assay, concentrations of indole from 10-3 – 10-9 M were chosen for the Western blot and permeability analysis.
Indole, which is produced in the intestine by some gut microbes and is the precursor of IS, significantly induced the expression of occludin and ZO-1 (Fig 3-12), which are essential components of epithelial and endothelial barriers (Feldman et al., 2005). However, the expression of the two proteins was significantly decreased when cells were incubated with 10-9 M indole compared to cells without indole treatment.
This result suggested that higher concentration of indole could maintain the intestinal barrier by inducing the expression of tight junction proteins.
The integrity of intestinal epithelium was performed by using TER measurements. Figure 3-13 demonstrated that the relative change in TER was increased
when cells were incubated with 10-3 M indole for 20 h. However, the relative change in TER of cells with 10-6 and 10-9 M indole treatment was similar to cells without indole treatment after 20 h incubation (Fig 3-13). This result suggested that higher concentration of indole could increase the integrity of the intestinal barrier. The results of Western blot and TER measurement implied that the increased intestinal barrier with higher concentration of indole may be related to the increased expression of tight junction proteins.
Figure 3-11. Effect of indole on the cell viability of Caco-2 cells assayed by MTT.
Control was cells cultured only with growth medium. Zero concentration means cells were only treated with N, N-Dimethylformamide (DMF). Values represent mean ± SD.
Error bars are obtained from three independently experiments. Statistical analysis was performed by one-way ANOVA followed by Dunnett’s multiple comparisons test.
Different letters indicated significant differences between groups (p < 0.05).
Control
0 10-9 10-6 10-3 10-2 0.0
0.2 0.4 0.6 0.8 1.0 1.2
Indole concentration (mol/L)
Ce ll vi abi lit y ( % )
a
b
a a
a
a
(A) (B)
Figure 3-12. Effect of indole on the expression of tight junction proteins (A) occludin and (B) 1 in Caco-2 cells. The levels of expression of occludin and ZO-1 were assessed by immunoblotting. Each lower subpanel showed an immunoblot and each upper subpanel showed the densitometric analysis of that blot. Values represent mean ± SD. Error bars are obtained from two independently experiments. Statistical analysis was performed by one-way ANOVA followed by Dunnett’s multiple comparisons test. Different letters indicated significant differences between groups (p
< 0.05). occludin /β-actin relative optical density
a
ZO-1/ β-actin relative optical density a
b c
Figure 3-13. Effect of indole on the transepithelial resistance of Caco-2 cells.
Changes in the TER of polarized Caco-2 cells exposed to DMF (control, filled diamonds), 10-9 M (open circles), 10-6 M (filled triangles), 10-3 M (filled circles) for 20 h. Values represent mean ± SD from three measurements at each time point and two independently experiments. Statistical analysis was performed by one-way ANOVA followed by Dunnett’s multiple comparisons test. Significant difference was set at p <
0.05.
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
0 4 8 12 16 20 24
Relative change in TER
Time (h)
Control
-9
-6
-3
10-9M 10-6M 10-3M
Discussion
Useful markers that can reflect the progression of CKD could not only help the vets to monitor the renal function of the cats, but also prevent them from moving to the severe stage of CKD. IS had been reported to be associated with the progression of CKD (Cheng et al., 2015). The concentration of IS detected in cats with IRIS stage 2 was significantly higher than that in control groups, which could be a useful marker for monitoring the progression of CKD. Administration of indole had been published to improve the progression of CKD in rats by producing the IS (Niwa et al., 1994). As a result, we tested the association between the concentration of indole in the feces and the concentration of IS in the blood. The results suggested that the concentration of indole was associated with the concentration of IS; however, the concentration of indole was not significant difference between IRIS stages, which suggested that IS, but not indole, could be a marker for predicting the progression of CKD.
More and more researchers had revealed a bi-directional relationship between CKD and gut microbiota. Alteration of gut microbiota in human patients and animal models with CKD had been reported (Jiang et al., 2017; Yoshifuji et al., 2016), and this is the first study that investigates the gut microbiota of cats with CKD. The composition of gut microbiota may be affected by many factors, and the living environment had been reported to be an important effect to alter the structure of gut microbiota (Kelsen and Wu, 2012). As a result, we collected CKD and corresponding cats from the same living condition to reduce the effect of environmental factors to the gut microbiota. Cats from four different family were recruited in this study. According to the principle coordinate analysis, there was a separation trend of controls and cats with CKD, but
cats from the same living conditions did not show a cluster (Fig. 3-3). This result implied that the effect of disease to the gut microbiota was more dominant than the effect of environmental factor in this study.
cats from the same living conditions did not show a cluster (Fig. 3-3). This result implied that the effect of disease to the gut microbiota was more dominant than the effect of environmental factor in this study.