Comparisons between CR and CZ flies

In CR flies, we detected 15–20 bacterial families common to all fly strains but the total number of bacterial families detected within each strain varied from 30 to 100 families (Figure 1A and S_obs index in Figure 1B). Of note, the number of cultivatable bacteria was much smaller than the total number of molecularly identified families (Table S1 and Table S2). For all fly strains, we observed a reduction in diversity with age coupled to Acetobacteraceae expansion (Figure 1, A and B). Lactobacillaceae, another dominant strain found in CR Drosophila, was also present in all four strains; however, not to the same extent as Acetobacteraceae. In addition, there was an age-dependent increase in bacterial load measured as indicated by CFUs across the life span of all four strains in accordance with previous literature (Wong et al. 2011; Figure S4A) as well as determined by molecular quantification of the V3 region of the 16S rRNA gene by QPCR across the life span of all four strains (Figure S4B).

In CZ flies, a statistically significant increase in gut microbiota diversity in all fly strains compared to their CR self was revealed (Figure 2 and Table S4 for diversity t-tests). For example, pairwise comparisons at day 20 between Shannon H and Simpson indices of diversity in CR and CZ conditions resulted in P-values between 0.0008 (pirk;trbd), 0.0003 (dif-key), and <0.0001 (yw, DGRP-208) including their biological (day 20) CR repeats. Specifically, Acetobacteraceae and to a lesser extent Lactobacillaceae were replaced by other bacterial families in the gut of yw and DGRP-208 (compare Figure 1, A and B; see below). Familial transmission of these bacteria therefore blocked diversity and made the gut microbiota structure simpler in CR conditions (see Table S4 for comparisons between CR and CZ). In contrast to yw and DGRP-208, Acetobacteraceae and Lactobacillaceae still remained an important part of the microbiota of the two mutant strains. Diversity patterns revealed that, in CZ flies, dif-key was the most diverse strain of the four (Figure 2B and Table S4 for statistics of pairwise comparisons). Pairwise comparisons for β diversity in both CR and CZ flies measured through χ² did not give meaningful results since all comparisons were found to be significant (Table S5). This indicated that the samples were overdispersed and instead, factorial β diversity tests were used to compare UniFrac analyses (Table S6).

Bacterial density in the gut of all young (5-day-old) CZ flies was significantly reduced in relation to their CR counterparts (Figure S4C). Moreover, during the time of observation CZ flies had consistently lower bacterial loads than their CR counterparts (Figure S4C). This showed that the main contributor of bacterial density in CR flies was familial transmission. However, a notable exception was yw CZ, which in 2 weeks obtained the same density levels as yw CR (Figure S4C).

Further analysis at the level of genus for the most abundant families revealed that, under CR conditions, the most representative bacterial genera were Acetobacter and Lactobacillus, while Acidimonas, Corynebacterium, and Shewanella followed at high levels (Figure S5). By removing competition from familial transmitted bacteria, genera such as Shewanella and those belonging to the family of Rhodobiaeae were able to thrive. Acetobacter and Lactobacillus remained detectable but at significantly lower levels (Figure S5).

Finally, culture-dependent experiments (Table S1) identified L. plantarum, L. pentosus, and A. pasteurianus, as the dominant species of Lactobacillaceae and Acetobacteraceae during the life span. As shown previously, L. plantarum has a stable association with the Drosophila gut (Storelli et al. 2011).

Comparisons in gut microbiota of wild-type and control strains

Relative abundance measurements (Figure 1A) coupled with Shannon H index analysis (Figure 1B) indicated that, between them, yw and DGRP-208 differed greatly in the diversity of their intestinal microbiota. Unlike the Simpson’s index, which is largely influenced by the dominant bacterial families, the Shannon H index considers all bacterial families equally and is affected by families with few representatives. Although both DGRP-208 and yw followed the same trends in their gut microbiota across the life span, DGRP-208 was a more diverse strain bearing a number of bacterial families (with low representation) that were not found in yw. Thus, the diversity of DGRP-208 across the life span was greater than that of yw (reflected in the t-test results of the Shannon H and Simpson index, see Table S4). For example, comparing diversity between DGRP-208 and yw, in CR conditions: day 2 Shannon H P-value = 0.0758, Simpson P-value = 0.0101; day 7 Shannon H P-value = 0.0738, Simpson P-value = 0.4274; day 7 Repeats Shannon H P-value = 0.0623, Simpson P-value = 0.4821; day 14 Shannon H P-value = 0.0738, Simpson P-value = 0.4274; day 20: Shannon H P-value = 0.2019, Simpson P-value = 0.4484; and day 20 Repeats: Shannon H P-value = 0.1242, Simpson P-value = 0.4272). This was in accordance with the increased number of bacterial families shown in S_obs (Figure 1B). However, on day 40 PE, both strains showed statistically indistinguishable α diversity indices (Shannon H P-value = 0.9447 and Simpson P-value = 0.8468), as was evident from the dominance of Acetobacter in their gut microbiota. Therefore, the α diversity analysis showed that DGRP-208 had a more diverse microbiota structure.

To compare dissimilarities between the microbiota of the two fly strains, we employed PCA (taking into account relative abundance, Figure 1C) as well as unweighted (taking into account phylogenetic relationship, Figure 1D) and weighted (taking into account both relative abundance and phylogenetic relationship, Figure 1E) PCoA. These analyses showed again that DGRP-208 and yw were very different at early ages (UniFrac distances for day 7, 0.8080 unweighted and 0.6098 weighted) but both converged in old age (UniFrac distances for day 40, 0.7473 unweighted and 0.4260 weighted). CFUs for both fly strains showed that they had the same levels of cultivable bacteria across the life span (Figure S4A). Nevertheless, there was an increase in bacterial load for both when compared to their 7-day self (Figure S4B). Moreover, QPCR analysis confirmed that these two fly strains followed the same trend/progression in bacterial load across their life span (Figure S4C).

In CZ flies, bacterial families that dominated the microbiota in CR conditions in both the yw and DGRP-208 strains, namely Acetobacteraceae and Lactobacillaceae, were replaced by others (Halomonadaceae and Shewanellaceae in DGRP-208 and Rhodobiaceae in yw) (Figure 2A). This strongly suggested that Acetobacteraceae and Lactobacillaceae were mainly present due to familial transmission. CZ yw flies presented an age-dependent reduction in diversity as seen in Simpson’s and Shannon’s H indices (Figure 2B). Nevertheless, both CZ fly strains were more diverse compared to their CR selves as seen in α diversity measurements of t-tests for both Shannon H and Simpson indices (Table S4). In addition, the statistically significant degree of dissimilarity in their CZ microbiota compared to their CR selves was shown in β diversity PCA (Figure 2C) and PCoA plots (Figure 2, D and E). There, CZ yw and DGRP-208 clustered away from their respective CR time points presenting greater UniFrac distances between these points than between CZ yw and CZ DGRP-208 (Table S4). For example, the unweighted UniFrac distance between 20 days CZ yw and its CR self was 0.9023, whereas between 20 days CZ yw and CZ DGRP-208 was 0.8277. The weighted UniFrac distance between 20 days CZ yw and its CR self was 0.7708, whereas between 20 days CZ yw and CZ DGRP-208 was 0.6211. Therefore, at 20 days, gut microbiota diversity was closer between CZ flies of DGRP-208 and yw than between CZ and CR flies of the same strain.

To compare the two fly strains in CZ conditions, we took into account relative abundance and α diversity t-tests of DGRP-208 and yw (see Table S4). We found that yw was consistently less diverse than DGRP-208 across the life span in CZ conditions (e.g., day 2 and day 2 repeats P < 0.0001 for both Shannon H and Simpson indices; day 20, P = 0.0007 for Shannon H and P < 0.0001 for Simpson), reflecting what was found in CR conditions and indicating an intrinsic difference in the two fly strains based on the genetic background.

Gut microbiota in immune-compromised flies

Dif-key CR guts displayed an early dominance of Lactobacillus that gave way to a gradual age-dependent increase in Acetobacteraceae, which following day 20 resembled a mono-association (>90%; Figure 1A). When each dif;key CR time point was compared to the corresponding yw CR time point, microbiota diversity of dif-key flies closely followed their genetic background across the early- and midlife span but differentiated from yw at day 40 (Figure 1B and Table S4). Specifically, α diversity t-tests for Shannon H and Simpson indices with respective P-values 0.7908 and 0.4493 for day 2; 0.1089 and 0.7273 for day 7; 0.2126 and 0.6528 for day 7 repeats, 0.8431 and 0.0190 for day 14; 0.3250 and 0.1263 for day 20; 0.3650 and 0.2165 for day 20 repeats; and <0.0001 for day 40. Interestingly, both 20- and 40-day-old dif-key CR sibling flies resembled each other indicating that their gut microbiota did not undergo any significant changes from middle age onwards (Figure 1, A–C). Unweighted β-diversity metrics (taking into account phylogenetic relationship) confirmed that the microbiota of 20- and 40-day-old dif-key CR sibling flies had the smallest UniFrac distance (0.7648) between each other than with any other time point/genotype (Figure 1D and Table S6). In their turn, both of these dif-key time points resembled the 20-day-old yw genetic background (Figure 1C). Weighted β diversity metrics (taking into account both relative abundance and phylogenetic relationship) indicated that 40-day-old dif;key flies had the smallest UniFrac distance (0.1854) with 20-day-old yw (Figure 1E and Table S6). This further underlined the lack of development with age in the microbiota structure of an immune-compromised gut, where Acetobacteraceae speedily took over the bacterial landscape. As expected, dif-key had a higher bacterial load compared to the yw control (Figure S4).

By removing maternally transmitted bacteria, relative abundance and diversity metrics (Simpson and Shannon indices) showed that the microbiota of CZ dif;key flies was more diverse than their yw background accompanied by a high number of distinct bacterial families (Figure 2B). Furthermore, the previous Acetobacteriaceae family domination was now replaced by bacteria belonging to the genus of Shewanella and genera representing the Halomonadaceae family (Figure 2A and Figure S5). Interestingly, following β diversity analysis (Figure 2C), microbiota patterns of dif-key CZ flies were found to be different from their genetic background. Specifically, unweighted UniFrac distances between CR yw and CR dif-key were 0.7633, 0.7829, and 0.7681 at 2, 14, and 20 days PE, respectively, whereas the same distances in CZ conditions were 0.7770, 0.8614, and 0.8888 (Table S4). Similar conclusions were drawn when weighted UniFrac distance were examined with distances between CR yw and CR dif-key at days 2, 14, and 20, i.e., 0.5504, 0.2995, and 0.3944, while in CZ conditions the same distances were 0.3999, 0.7976, and 0.5822 (Table S4). This revealed that familial transmission was dominant over the lack of immunity to shape the gut microbiota in CR conditions. When familial transmission was removed, the influence of a deficient immune system was highlighted (see Discussion).

Gut microbiota in a constitutively active immune system

Bacterial gut populations in pirk;trbd had a different dynamic across the life span compared to its genetic background, as indicated by relative abundance measurements (Figure 1A).

Diversity indices of CR guts showed that pirk;trbd flies started with a diversity comparable to all other fly strains (Figure 1B). However, at day 7, a consistent pattern of low diversity ensued. This pattern continued to middle age (day 14) where pirk;trbd gut microbiota were dominated by two bacterial families, namely Corynebacteriaceae and Acetobacteraceae, covering >95% of relative abundance. This resulted in low diversity compared to both yw and dif-key (see Simpson and Shannon H indices in Figure 1B and Table S4). On day 20 PE (very near the end of the strain’s median life span), the dominant gut OTU was Enterococcaceae (Figure 1A), represented mainly by bacteria belonging to the genus of Enterococcus (Figure S5).

Overall, the intestinal microbiota of both CR (Figure 1C) and CZ (Figure 2C) pirk;trbd flies were significantly different in terms of structure from their genetic background (see below). In addition, microbiota composition of pirk;trbd CR flies was also different compared with dif-key (Figure 1C). This indicated an effect of their immune status over and above familial transmission. To strengthen this conclusion, we applied factors-explained separation tests (adonis, anosim, and permanova tests) for both weighted and unweighted UniFrac distances. We grouped all CR pirk;trbd samples and compared them against all CR yw samples, asking if hyperactive immunity resulted in statistically distinguishable gut microbiota compared with control flies. Results gave significant differences for all tests applied (adonis test P-value = 0.03992, anosim test P-value = 0.06205, and permanova test P-value = 0.05433). In other words: all three tests showed significant probability of nonrandom separation of unweighted UniFrac distances (Figure 1D and Table S6). However, this was not the case when comparing dif-key vs. yw or DGRP-208 vs. yw (Figure 1D and Table S6). Moreover, using CR biological repeats (Figure S2) we obtained similar results (adonis P-value = 0.02842, anosim P-value = 0.052877, and permanova P-value = 0.05821), which confirmed the role of hyperactive immunity, and the consistency of the results across different generations and over any seasonal fluctuations of microbiota (Wong et al. 2011). The statistically significant differences between CR pirk;trbd and yw revealed in unweighted UniFrac distances were not presented when weighted UniFrac distance were measured (Figure 1E and Table S6). The latter analysis takes into account relative abundance not computing equally, therefore rare and major OTUs. This meant that differences observed in the unweighted UniFrac analysis were mostly due to differences in minor components of the community (see Discussion). In addition, pirk;trbd CR and CZ had a variable but consistently high bacterial load across the life span, in contrast to all other strains where there was an age-dependent increase of bacterial load (Figure S4).

Taken together, our results indicated that flies with a constitutively active immune system had a structure and density of gut microbiota that differed both from their genetic background and from immune-compromised flies. Moreover, persistence of Acetobacteraceae in CR pirk;trbd was also observed in CZ pirk;trbd. Therefore, this bacterial family was not outcompeted by other bacteria in CZ conditions as seen in the other fly strains (Figure 2A). Thus, constitutively active immunity may result in the acceleration of the Acetobacteraceae-dominated ageing pattern in microbiota composition.

Pirk-trbd CZ flies showed diversity that was statistically indistinguishable from dif-key CZ and yw CZ flies at young age (2 days PE; Figure 2B). Specifically, respective P-values of t-tests for Shannon H and Simpson indices between pirk;trbd and yw were 0.5084 and 0.7315 (0.5986 and 0.6835 for the biological repeats) (Table S4). Previously present bacterial families in CR pirk;trbd were replaced by bacterial genera belonging to the families of Halomonadaceae and Swevanellaceae following similar trends as CZ dif-key flies (Figure 2A and Figure S5). Fourteen-day-old pirk;trbd CZ flies showed lower diversity than both dif-key and yw flies represented by Acetobacteraceae domination (Figure 2, A and B and Table S4). This was much like the expansion seen in pirk;trbd CR (Figure 1A). However, at the age of 20 days PE, pirk;trbd guts tolerated this Acetobacter domination and gained in diversity (Figure 2A and Table S4). PCA plots (see Figure 2C) showed that, as they aged, the degree of dissimilarity between CZ pirk;trbd and CZ dif-key was less than the dissimilarity between each one of them and their yw genetic background. For example, unweighted UniFrac distance between CZ pirk;trbd and CZ dif-key at day 20 was 0.7496, while the same distance between CZ pirk;trbd and CZ yw was 0.8792 and between CZ dif-key and CZ yw was 0.8355 (Figure 2D and Table S6). Weighted UniFrac distances confirmed this with the distance between CZ pirk;trbd and CZ dif-key at day 20, at 0.3395; the same distance between CZ pirk;trbd and CZ yw was 0.7980 and between CZ dif-key and CZ yw was 0.5822 (Figure 2E and Table S6). Therefore, when familial transmission was eliminated, middle-aged flies from immune-aberrant strains resembled each other more than their genetic background. Moreover, removing familial transmission extensively changed the age-dependent dynamics of the gut microbiota structure and diversity in pirk;trbd, indicating that familial transmission strongly interacted with immune status in defining the bacterial landscape of the gut in these flies. Nevertheless, CZ pirk;trbd flies were more similar to their CR self than CZ yw (unlike DGRP-208 and yw for example). Specifically, PCA plots showed that both yw CR and CZ flies clustered away from pirk;trbd CZ (Figure 2C). In addition, weighted UniFrac distances between CZ pirk;trbd flies and their CR self was 0.6977 while the one between CZ pirk;trabid and CZ yw flies was 0.7980 (Figure 2E and Table S6). Therefore, even when familial transmission was removed, the significant differences between pirk;trbd and the yw genetic background remained.

Immune status and gut physiology

Since microbial colonization depends on the physical and chemical conditions of the different gut compartments, we checked several factors involved in gut physiology. Applying two different dyes (BCP, purple at pH 6.8 and yellow at pH 5.2 and BPB, yellow at pH < 3 and blue at pH > 4.6), we detected no differences in adult gut pH at various points across the life span (7, 20, and 40 days PE) (Figure S6 and Figure S7). Additional features including gut length (Figure S8A) and gut internuclei distance (foregut, midgut, and hindgut) (Figure S8, B–D, respectively) were checked in all fours CR strains and again showed no difference.

Predictably, AMP gene expression levels were markedly increased in pirk;trbd flies compared to yw throughout their life span, despite the fact that there was an age-dependent increase in AMP transcription in all fly strains (Figure 3A). This indicated an inflamed gut with stressed enterocytes, which in turn implied increased stem cell activity. Indeed, we found a statistically significant increase of early (2 and 7 days) gene expression in pirk;trbd for the cytokine upd3, which triggers Janus kinase signal transducer and activator of transcription (JAK-STAT) signaling in stem cells (Figure 3B). This reflected an early increase of JAK-STAT activity measured by gene expression of the JAK-STAT pathway target socs36E (Figure 3C). Furthermore, there was a statistically significant difference in enterocyte cell death in pirk;trabid guts, as detected by the vital dye Sytox green across the life span in both larvae and adults (Figure 4A for images and Figure 4B for quantification). Increased AMP levels and high cell death was accompanied by an increase in bacterial load across the life span of pirk;trbd (Figure S4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Immune status and JAK-STAT activity. (A) mRNA gene expression of three AMPs (attacin A, attacin D, and diptericin B) with age in the four fly strains (DGRP-208, yw, dif-key, and pirk;trbd). Error bars represent the SD of five separate experiments; ns, P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 indicate significance value when compared to yw day 7. (B) Upd3 mRNA gene expression in all four strains across life span. Error bars represent the SD of three separate experiments; ns, P > 0.05, *P < 0.05, **P < 0.01, and ***P < 0.001 indicate significance value when compared to yw day 2. (C) Socs36E mRNA gene expression in four strains with age. Error bars represent the SD of three separate experiments; ns, P > 0.05, *P < 0.05 specify significance value when compared to yw day 2. ns, not significant.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

The effect of cohousing on gut microbiota composition. (A) The relative abundance of bacterial families detected in dissected guts of cohoused strains (yw and dif-key; yw and pirk;trbd) disclosed by 16S rRNA sequencing. (B) α diversity indices (Simpson’s and Shannon H indices) and total number of observed families (S_obs) for the four strains at selected time points (day 2 and 14). (C) Principal Component Analysis (PCA) of the bacterial community of cohoused strains and corresponding CZ strains. (D) Principal Coordinate Analysis (PCoA) of unweighted UniFrac distances of the cohoused strains and corresponding strains in CZ conditions. (E) PCoA of weighted UniFrac distances. CR, conventionally reared; CZ, conditionalized; ns, not significant.

All CZ strains exhibited reduced cell death in comparison to CR guts (see Figure 4A for images and Figure 4B for quantification). Specifically, CZ conditions lowered gut cell death in very low levels in larvae while the age-dependent increase in gut cell death observed in CR adults was also significantly reduced (Figure 4B). This showed the importance of the microbiota in regulating gut homeostasis and revealed its major effect in cases of unregulated immunity (as in pirk;trbd), where inflamed conditions were promoted by the presence of bacteria in the gut.

Constitutive immunity influences the local bacterial microenvironment

Familial transmission and food are the sources of bacteria for laboratory cultures of Drosophila (Wong et al. 2015). The contribution of familial transmission can be addressed by producing CZ flies. However, as recently shown, the host drives microbiota changes there through its constant interaction with food, which in turn alters the gut microbiota landscape (Wong et al. 2015). The microenvironment therefore plays an important role in shaping gut microbiota. But what is the influence of the immune status of flies on the microenvironment?

To find out whether microbiota differences in CZ flies could be influenced by immune status during co-culture, we mixed bleached fertilized eggs of pirk-trbd or dif-key flies with yw. Using high-throughput NGS, we determined the bacterial communities in guts after 2 weeks of co-culture (14 days PE) and compared them to CZ flies at 2 and 14 days of single culture (see Figure 5A). Of note, that crowding conditions in both single and co-culture were identical with a total of 30 flies in each vial. Dif-key flies retained constant levels of diversity and exhibited the same pattern of age-dependent increase of diversity as single cultured dif-key CZ flies (Figure 5B; see Simpson and Shannon H indices). Likewise, yw remained more comparable to their single culture than to dif-key when co-cultured with the latter (Figure 5C). PCoA plots showed that unweighted as well as weighted UniFrac distances were smaller between yw CZ and yw cohoused with dif-key than between yw cohoused with dif-key and their dif-key cohabitants (Figure 5, D and E and Table S6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Enterocyte cell death in CR and CZ dissected guts. (A) Sytox green staining of the dissected guts from 5 day larvae, 2 and 20 day adults of the four strains (DGRP-208, yw, dif-key, and pirk;trbd) under CR and CZ conditions. (B) The number of dead cells of the above four strains at the above time points under CR and CZ conditions. Error bars represent SD of six separate experiments; ns, P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 indicate significant values when compared to yw of the corresponding time point. CR, conventionally reared; CZ, conditionalized; ns, not significant.

In contrast, co-culturing yw and pirk;trbd resulted in gut microbiota of yw being driven by pirk;trbd toward an almost mono-association with Acetobacteraceae and a complete absence of Lactobacillaceae (Figure 5A). Moreover, yw flies exhibited reduced diversity with age, thus resembling pirk;trbd (Figure 5B; see Simpson and Shannon H indices and compare to Figure 1B and Figure 2B). In addition, PCA indicated that by the end of co-culturing (14 days), the gut microbiota landscape of yw became more like their nonsibling pirk;trbd cohabitants (Figure 4C). PCoA plots showed that both unweighted and weighted UniFrac distances between pirk;trbd and yw cohabitants were less than distances of cohoused yw with their siblings in single culture (Figure 5, D and E and Table S1). Specifically, at 14 days, the UniFrac distances between yw and their pirk;trbd cohabitants was 0.1992 (weighted) and 0.7129 (unweighted), whereas with their yw CZ siblings (single culture) UniFrac distances were 0.6091 (weighted) and 0.7965 (unweighted). Taken together, these results showed that pirk;trbd had a statistically significant influence on yw microbiota through their common microenvironment.