Study population
This study included 156 subjects (Table 1) comprising 129 (82.7%) HIV-1 infected patients with different clinical profiles and 27 (17.3%) negative controls recruited in Barcelona, Spain, between January and December 2014. HIV-1 infected subjects were enrolled from two tertiary HIV-1 clinics and negative controls were recruited from a cohort of HIV-negative MSM at risk of becoming infected by HIV-1 attending a community centre and HIV-1-negative partners from HIV-1-infected subjects attending the HIV clinics19,21,22. Mean age of included patients was 43 years, most were male (79%) and of Caucasian ethnicity (79%). Patients were classified according to HIV-1 infection risk group in MSM (n=100) and no-MSM (n=56), according to faecal microbiome cluster in Bacteroides (n=63) or Prevotella (n=93) enriched enterotypes19 and according to microbial gene richness values obtained by whole faecal metagenome shotgun sequencing in high-gene count (HGC) (n=53) or low-gene count (LGC) (n=103)18. Low microbial gene counts have been previously linked to gut dysbiosis in different gut inflammatory diseases23. Additionally, in a previous study conducted in the same cohort of patients a significant and independent dose-effect association between nadir CD4+ T-cell counts and LGC was identified18. Most MSM showed a Prevotella enriched enterotype (88%) and no-MSM a Bacteroides enriched enterotype (91%) as previously described19. Regarding gene richness, most no-MSM subjects presented gut microbiome with LGC (88%). Subjects in the no-MSM group were older and showed lower Nadir CD4+ T-cell counts compared to MSMs.
Table 1 Patient's characteristics according to sexual preference and gene richness
|
Overall, N = 1561
|
|
Sexual preference
|
|
Gene Richness
|
|
MSM, N = 1001
|
no MSM, N = 561
|
p-value2
|
|
HGC, N = 531
|
LGC, N = 1031
|
p-value2
|
Age
|
43 (35-51)
|
|
38 (34-46)
|
50 (42-54)
|
<0.001
|
|
38 (35-46)
|
46 (36-53)
|
0.024
|
Gender
|
|
|
|
<0.001
|
|
|
|
0.018
|
Women
|
31 (20)
|
|
0 (0)
|
31 (55)
|
|
|
5 (9.4)
|
26 (25)
|
|
Men
|
124 (79)
|
|
99 (99)
|
25 (45)
|
|
|
47 (89)
|
77 (75)
|
|
Tansgender women
|
1 (0.6)
|
|
1 (1.0)
|
0 (0)
|
|
|
1 (1.9)
|
0 (0)
|
|
Ethnicity
|
|
|
|
0.2
|
|
|
|
0.005
|
Asiatic
|
1 (0.6)
|
|
0 (0)
|
1 (1.8)
|
|
|
1 (1.9)
|
0 (0)
|
|
Caucasian
|
124 (79)
|
|
78 (78)
|
46 (82)
|
|
|
36 (68)
|
88 (85)
|
|
Hispanic-Latin
|
28 (18)
|
|
21 (21)
|
7 (13)
|
|
|
16 (30)
|
12 (12)
|
|
Other
|
3 (1.9)
|
|
1 (1.0)
|
2 (3.6)
|
|
|
0 (0)
|
3 (2.9)
|
|
BMI
|
23.8 (22.0-26.1)
|
|
24.3 (22.3-26.2)
|
23.5 (20.9-25.2)
|
0.053
|
|
24.4 (22.3-26.3)
|
23.7 (21.8-25.5)
|
0.2
|
Missing values
|
18
|
|
16
|
2
|
|
|
7
|
11
|
|
HIV-1 status
|
|
|
|
0.012
|
|
|
|
0.002
|
Negative
|
27 (17)
|
|
23 (23)
|
4 (7.1)
|
|
|
16 (30)
|
11 (11)
|
|
Positive
|
129 (83)
|
|
77 (77)
|
52 (93)
|
|
|
37 (70)
|
92 (89)
|
|
HIV-1 phenotype
|
|
|
|
|
|
|
|
|
Concordant
|
53 (34)
|
|
28 (28)
|
25 (45)
|
|
|
11 (21)
|
42 (41)
|
|
Discordant
|
18 (12)
|
|
6 (6.0)
|
12 (21)
|
|
|
3 (5.7)
|
15 (15)
|
|
Early-treated
|
13 (8.3)
|
|
12 (12)
|
1 (1.8)
|
|
|
5 (9.4)
|
8 (7.8)
|
|
Elite controller
|
8 (5.1)
|
|
3 (3.0)
|
5 (8.9)
|
|
|
3 (5.7)
|
5 (4.9)
|
|
Late presenter
|
11 (7.1)
|
|
8 (8.0)
|
3 (5.4)
|
|
|
2 (3.8)
|
9 (8.7)
|
|
ART-naive
|
15 (9.6)
|
|
13 (13)
|
2 (3.6)
|
|
|
7 (13)
|
8 (7.8)
|
|
Viremic controller
|
11 (7.1)
|
|
7 (7.0)
|
4 (7.1)
|
|
|
6 (11)
|
5 (4.9)
|
|
HIV-1 negative
|
27 (17)
|
|
23 (23)
|
4 (7.1)
|
|
|
16 (30)
|
11 (11)
|
|
Antiretroviral treatment
|
66 (42)
|
|
40 (40)
|
26 (46)
|
0.4
|
|
16 (30)
|
50 (49)
|
0.028
|
Gene richness
|
|
|
|
<0.001
|
|
|
|
|
HGC
|
53 (34)
|
|
46 (46)
|
7 (13)
|
|
|
|
|
|
LGC
|
103 (66)
|
|
54 (54)
|
49 (88)
|
|
|
|
|
|
HIV-1 risk group
|
|
|
|
|
|
|
|
<0.001
|
MSM
|
100 (64)
|
|
|
|
|
46 (87)
|
54 (52)
|
|
no MSM
|
56 (36)
|
|
|
|
|
7 (13)
|
49 (48)
|
|
Microbiome cluster
|
|
|
|
<0.001
|
|
|
|
<0.001
|
Bacteroides
|
63 (40)
|
|
12 (12)
|
51 (91)
|
|
|
7 (13)
|
56 (54)
|
|
Prevotella
|
93 (60)
|
|
88 (88)
|
5 (8.9)
|
|
|
46 (87)
|
47 (46)
|
|
Antibiotic intake, previous 3 months
|
2 (1.3)
|
|
2 (2.0)
|
0 (0)
|
0.5
|
|
0 (0)
|
2 (1.9)
|
0.5
|
Antibiotic intake, previous 6 months
|
35 (22)
|
|
20 (20)
|
15 (27)
|
0.3
|
|
9 (17)
|
26 (25)
|
0.2
|
HIV-1 RNA, copies/mL3
|
|
|
0.044
|
|
|
|
0.034
|
Undetectable
|
85 (66)
|
|
45 (58)
|
40 (78)
|
|
|
19 (51)
|
66 (73)
|
|
<=10.000
|
22 (17)
|
|
15 (19)
|
7 (14)
|
|
|
11 (30)
|
11 (12)
|
|
> 10.000
|
21 (16)
|
|
17 (22)
|
4 (7.8)
|
|
|
7 (19)
|
14 (15)
|
|
Missing values
|
1
|
|
0
|
1
|
|
|
0
|
1
|
|
CD4+ T-cell counts3
|
705 (469-856)
|
|
727 (490-851)
|
636 (288-934)
|
0.8
|
|
772 (570-860)
|
644 (289-853)
|
0.11
|
Missing values
|
1
|
|
1
|
0
|
|
|
0
|
1
|
|
Nadir CD4+ T-cell counts3
|
337 (140-529)
|
|
372 (209-577)
|
244 (91-438)
|
0.005
|
|
443 (339-601)
|
280 (113-492)
|
0.002
|
Missing values
|
2
|
|
2
|
0
|
|
|
1
|
1
|
|
CD8+ T-cell counts3
|
777 (576-1,012)
|
|
779 (627-983)
|
777 (478-1,027)
|
0.6
|
|
749 (604-1,158)
|
792 (559-991)
|
0.3
|
Missing values
|
1
|
|
1
|
0
|
|
|
1
|
0
|
|
CD4+/CD8+ ratio3
|
0.84 (0.52-1.22)
|
|
0.83 (0.55-1.19)
|
0.92 (0.46-1.34)
|
0.7
|
|
0.81 (0.55-1.16)
|
0.88 (0.49-1.32)
|
>0.9
|
Missing values
|
2
|
|
2
|
0
|
|
|
1
|
1
|
|
1 n(%); median (IQR)
|
2 Wilcoxon rank sum test; Fisher’s exact test; Pearson’s Chi-squared test
|
3 Values obtained only for HIV-1 positive subjects (n=129)
|
Gut resistome diversity
A total of 308 different AMRD grouped in 97 antimicrobial resistant (AMR) gene families were identified in the overall analysed samples. The most abundant AMR gene families in this study were tetracycline-resistant ribosomal protection protein, CfxA beta-lactamase, 23S rRNA with mutation conferring resistance to macrolide antibiotics, 16S rRNA with mutation conferring resistance to aminoglycoside antibiotics and Erm 23S ribosomal RNA methyltransferase conferring resistance to macrolide, lincosamide and streptogramin (MLS) antibiotics.
No differences in gut resistome alpha diversity and composition were identified according to HIV-1 infection status, HIV-1 profile or whether subjects had initiated antiretroviral treatment or had previously taken antibiotics at the time of inclusion. Additionally, we did not identify significant correlations between gut resistome alpha diversity and CD4+ T-cell counts, nadir CD4+ T-cell counts, CD8+ T-cell counts and CD4+/CD8+ ratio. However, we identified a significantly more diverse and a tendency towards a richer gut resistome in MSM compared to no-MSM subjects (Figure 1A). The same differences were observed when comparing Prevotella and Bacteroides enriched enterotypes (sFigure 1). Regarding gene richness, a significantly higher alpha resistome diversity was identified in HGC compared to LGC microbiomes (Figure 1B).
When analysing gut resistome composition, beta-diversity analyses showed significantly different resistome composition according to sexual preference (R2=0.1, p-value=0.001), microbiome cluster (R2=0.1, p-value=0.001) and gene richness (R2=0.06, p-value=0.001) (Figure 2, sFigure2). Initially, a univariate PERMANOVA analysis was conducted identifying a set of significant variables which were included in the multivariate PERMANOVA analysis. The multivariate analysis showed that sexual preference, microbiome cluster, HIV-1 phenotype and microbiome gene richness remained independently significant contributing to differences in gut resistome composition (Table 2). Interestingly, in the multivariate PERMANOVA analysis there was an R2 value decrease of the microbiome cluster variable when combined with sexual preference, demonstrating a high correlation between both variables. When pairwise comparisons were performed according to the different HIV-1 phenotypes, only Discordant versus ART-naïve and HIV-1 negative versus Elite controller comparisons remained significant (sTable 2).
Table 2 Univariate and multivariate PERMANOVA analysis
|
|
Univariate analysis
|
|
Multivariate analysis
|
|
Multivariate analysis (missing values)
|
p-value
|
R2 value
|
p-value
|
R2 value
|
p-value
|
R2 value
|
HIV-1 risk group
|
|
0,001
|
0,1
|
|
0,001
|
0,1
|
|
0,001
|
0,1
|
Microbiome cluster
|
|
0,001
|
0,1
|
|
0,001
|
0,02
|
|
0,001
|
0,02
|
HIV-1 phenotype
|
|
0,001
|
0,08
|
|
0,011
|
0,05
|
|
0,007
|
0,06
|
Gene richness
|
|
0,001
|
0,06
|
|
0,001
|
0,05
|
|
0,001
|
0,04
|
Gender
|
|
0,001
|
0,05
|
|
0,706
|
0,01
|
|
0,489
|
0,01
|
Age
|
|
0,001
|
0,03
|
|
0,116
|
0,01
|
|
0,11
|
0,01
|
Nadir CD4+ T-cell counts
|
|
0,001
|
0,02
|
|
-
|
-
|
|
0,991
|
0
|
HIV-1 RNA, copies/mL
|
|
0,04
|
0,02
|
|
-
|
-
|
|
-
|
-
|
Ethnicity
|
|
0,042
|
0,03
|
|
-
|
-
|
|
-
|
-
|
Antibiotic intake, previous 3 months
|
|
0,052
|
0,01
|
|
-
|
-
|
|
-
|
-
|
Antibiotic intake, previous 6 months
|
|
0,182
|
0,01
|
|
-
|
-
|
|
-
|
-
|
BMI
|
|
0,233
|
0,01
|
|
-
|
-
|
|
-
|
-
|
HIV-1 status
|
|
0,271
|
0,01
|
|
-
|
-
|
|
-
|
-
|
CD8+ T-cell counts
|
|
0,443
|
0,01
|
|
-
|
-
|
|
-
|
-
|
Antiretroviral treatment
|
|
0,507
|
0,01
|
|
|
|
|
|
CD4+ T-cell counts
|
|
0,561
|
0,01
|
|
-
|
-
|
|
-
|
-
|
CD4+/CD8+ ratio
|
|
0,693
|
0,01
|
|
-
|
-
|
|
-
|
-
|
Differentially abundant antibiotic resistance determinants
We evaluated differentially abundant AMRD according to sexual preference, microbiome gene richness (Table 3) and microbiome cluster (sTable 3) identifying a set of significantly enriched determinants.
We identified that MSM microbiome were enriched in 16S rRNA with mutations conferring resistance to aminoglycoside antibiotics, 23S rRNA with mutations conferring resistance to macrolide and pleuromutilin antibiotics, ANT (6), enzyme conferring resistance to aminoglycosides and CfxA and ACl beta-lactamases conferring resistance to cephalosporin and cephamycin antibiotics, respectively. On the other hand, MSM were depleted in CblA beta-lactamase conferring resistance to cephalosporins, sulfonamide resistant sul, ABC-F ATP-binding cassette ribosomal protection protein and major facilitator superfamily (MFS) antibiotic efflux pump conferring resistance to different antibiotic classes, 23S rRNA with mutation conferring resistance to streptogramins antibiotics and tetracycline-resistant ribosomal protection protein (Table 3). Most AMRD enriched in MSM and no-MSM groups were also significantly enriched in Prevotella and Bacteroides enterotypes, respectively (sTable 3).
According to gene richness CblA and CfxA beta-lactamases, ABC-F ATP-binding cassette ribosomal protection protein and APH (3') resistance determinants were enriched in LGC group. Resistance determinants enriched in HGC group were 23S rRNA with mutation conferring resistance to macrolide antibiotics, 16S rRNA with mutation conferring resistance to tetracycline derivatives, ACI beta-lactamase and chloramphenicol acetyltransferase (CAT).
Not surprisingly, a set of differentially abundant AMRD were among determinants with higher loadings on ordination components in the resistome beta-diversity composition (Figure 3, sFigure 3).
Namely, CblA beta-lactamase, ABC-F ATP-binding cassette ribosomal protection protein (both AMRD enriched in no-MSM, Bacteroides and LGC groups) and MSF antibiotic efflux pump (enriched in no-MSM and Bacteroides groups) showed negative significant loading towards NMDS component 1. On the other hand, 23S rRNA with mutation conferring resistance to macrolide antibiotics (enriched in MSM, Prevotella and HGC groups) showed a positive significant loading towards NMDS component 1. Finally, CfxA beta-lactamase (enriched in MSM, Prevotella and LGC groups) showed a positive significant loading towards NMDS component 2. As expected, the loadings showed a correlation with the grouping variables in which the different AMRD were enriched (Figure 3, sFigure 3).
We did not identify a clear tendency towards antibiotic classes the identified AMRD were conferring resistance to and sexual preference, gene richness or microbiome cluster.
Table 3 Differentially abundant antimicrobial resistance gene families according to Sexual preference and Gene richness
Antimicrobial resistance gene family
|
Sexual preference
|
|
Gene Richness
|
Drug classes
|
MSM1
|
no MSM1
|
adjusted p-value
|
log2FC
|
Group
|
|
HGC1
|
LGC1
|
adjusted p-value
|
log2FC
|
Group
|
|
CblA beta-lactamase
|
0
|
8111
|
0,00
|
-3,58
|
no MSM
|
|
0
|
2333
|
0,00
|
-2,55
|
LGC
|
Cephalosporin
|
Sulfonamide resistant sul
|
0
|
638
|
0,00
|
-3,24
|
no MSM
|
|
-
|
-
|
-
|
-
|
-
|
Sulfonamides
|
ABC-F ATP-binding cassette ribosomal protection protein
|
1362
|
9316
|
0,00
|
-2,31
|
no MSM
|
|
1235
|
4203
|
0,00
|
-1,63
|
LGC
|
Lincosamide, Macrolide, Oxazolidinone, Phenicol, Pleuromutilin, Streptogramin, Tetracycline
|
Major facilitator superfamily (MFS) antibiotic efflux pump
|
4713
|
25049
|
0,00
|
-1,83
|
no MSM
|
|
-
|
-
|
-
|
-
|
-
|
Macrolide, Tetracycline, Fluoroquinolone, Nucleoside, Aminoglycoside, Cephalosporin, Peptide, Rifamycin, Carbapenem, Penam, Fosfomycin, Lincosamide, Phenicol, acridine dye, disinfecting agents and intercalating dyes, Benzalkonium, Chloride, Rhodamine,
|
Lincosamide nucleotidyltransferase (LNU)
|
1767
|
4876
|
0,01
|
-1,20
|
no MSM
|
|
-
|
-
|
-
|
-
|
-
|
Lincosamide
|
23S rRNA with mutation conferring resistance to streptogramins antibiotics
|
23209
|
28260
|
0,04
|
-0,22
|
no MSM
|
|
-
|
-
|
-
|
-
|
-
|
Streptogramin
|
tetracycline-resistant ribosomal protection protein
|
102817
|
120759
|
0,02
|
-0,16
|
no MSM
|
|
-
|
-
|
-
|
-
|
-
|
Tetracyclines
|
16s rRNA with mutation conferring resistance to aminoglycoside antibiotics
|
71228
|
59609
|
0,03
|
0,27
|
MSM
|
|
-
|
-
|
-
|
-
|
-
|
Aminoglycosides
|
23S rRNA with mutation conferring resistance to macrolide antibiotics
|
83582
|
66531
|
0,00
|
0,32
|
MSM
|
|
88954
|
71935
|
0,00
|
0,30
|
HGC
|
Macrolides
|
ANT (6)
|
8626
|
3207
|
0,03
|
0,34
|
MSM
|
|
-
|
-
|
-
|
-
|
-
|
Aminoglycosides
|
23S rRNA with mutation conferring resistance to pleuromutilin antibiotics
|
15258
|
12043
|
0,00
|
0,38
|
MSM
|
|
-
|
-
|
-
|
-
|
-
|
Pleuromutilin
|
CfxA beta-lactamase
|
90580
|
59552
|
0,01
|
0,44
|
MSM
|
|
61481
|
90592
|
0,04
|
-0,48
|
LGC
|
Cephamycins
|
ACI beta-lactamase
|
1920
|
0
|
0,00
|
3,28
|
MSM
|
|
1819
|
625
|
0,05
|
0,12
|
HGC
|
Cephalosporins, Penams
|
16S rRNA with mutation conferring resistance to tetracycline derivatives
|
-
|
-
|
-
|
-
|
-
|
|
4817
|
1039
|
0,03
|
1,17
|
HGC
|
Tetracyclines
|
Chloramphenicol acetyltransferase (CAT)
|
-
|
-
|
-
|
-
|
-
|
|
2939
|
415
|
0,00
|
1,12
|
HGC
|
Phenicol
|
APH (3')
|
-
|
-
|
-
|
-
|
-
|
|
4206
|
2192
|
0,02
|
-0,11
|
LGC
|
Aminoglycosides
|
1 median RPKM (Reads Per Kilobase per Million mapped reads)
|