The construction of porcine skin explant wells wound model for biofilm development
The ex vivo biofilm formation by P. aeruginosa PAO1 in porcine skin explant wells was monitored using fluorescent microscopic imaging and plating. The fluorescent imaging showed that the PAO1 strain formed micro-colonies for 24 h in explant wells, developed mushroom-like structure at 48 h, and at 72 h mushroom-like structure disappeared, remaining a thin bacterial lawn (Fig. 2). The results of plating determined that when incubating for 48 h, the count of bacterial in explant wells is the most than incubating 24 h and 72 h, which is consistent with the results of imaging (Fig. S1). Under the tested conditions, it was found that within 72 h, the PAO1 went through the whole biofilm developing cycle of “initial attachment, micro-colonies, mature biofilm, and dispersal” (Mihai et al. 2019). To assess potential contamination from other microorganisms, the porcine skin explant wells incubated with LB medium were imaged and plated after 24 h, 48 h, and 72 h, respectively. The results showed that just a few contaminations were found after 72 h. However, no contamination was found when the porcine skin explant wells incubated with the PAO1 culture after 72 h, which implied the model was suitable for studying P. aeruginosa biofilm development in chronic wounds.
Transcriptome analysis at genome-wide level
In order to identify genes that are specifically expressed in biofilms, RNA-seq technique was used to determine the transcriptome of P. aeruginosa PAO1 for mature biofilm and planktonic cells. The total number of mapped reads ranged between 35,183,947 and 37,580,850 for biofilm samples and between 36,481,656 and 39,427,607 for planktonic samples. 88.57% of total reads were successfully aligned to the reference genome, and 82.77% of all mapped reads were aligned to annotated gene regions for biofilm samples. For planktonic samples, 87.07% of mapped reads were aligned to annotated gene regions (Table S1). In addition, blast alignment showed that the average rRNA contamination ratios were 4.88% and 3.62% for biofilm and planktonic samples, respectively (Table S1), which indicated the data could be used for further analysis. Furthermore, RNA-seq data analysis showed that a total of 5,405 (97.00%) and 5,296 (95.05%) genes were confidently identified in biofilm and planktonic cells, respectively. Of which, 5,275 genes were both expressed in biofilm and planktonic cells. Above data analysis further determined that the ex vivo wound model was suitable for studying P. aeruginosa biofilms developed in chronic wounds without contaminations. Notably, in these expressed genes, 130 genes uniquely expressed in biofilm cells and 21 genes were found exclusively in planktonic cells. However, the expression levels of these genes that were only detected in either planktonic or biofilm cells were very low except that transcripts associated with the type II hxc secretion system expressed in biofilm state, but their transcription levels were not very high. (Table S2 and Table S3).
The combined expression levels of all genes were determined in biofilm and planktonic samples. The relative mean expression values (RME) for each identical gene product were produced and the top 30 most highly expressed genes in biofilms were ranked (Fig. 3a). The corresponding values in planktonic samples are also displayed together in Fig. 3a. Similarly, the top 30 most highly expressed genes in planktonic samples were ranked (Fig. 3b). In the biofilm cells, these genes, encoding alkaline phosphatase L, hypothetical protein, heat-shock protein LbpA, and glutamine synthetase, were among the most highly expressed genes with normalized matrix count value more than 5,000. Moreover, nine genes were enriched to biological process with cellular protein metabolic pathway (FDR < 0.001). However, among the 30 highest abundant genes in planktonic cells, 20 genes were enriched to translation process and ribosome pathway, which determined active metabolism in planktonic cells. A comparison of the 30 highest abundant genes in biofilm and planktonic cells resulted in an overlap of 9 genes. Of which, genes encoding cell division protein MraZ and ribosome were the most enriched. Significantly, the expression levels of 21 genes uniquely highly expressed in biofilm were very low in planktonic cells, especially the expression levels of lapA encoding alkaline phosphatase and pstS encoding phosphate ABC binding protein (Fig. 3c, Table S4). In addition, 21 genes uniquely highly expressed in planktonic cells were related to ribosome. The expression levels of these genes were also high in biofilm cells, although they were not the top 30 highest expressed genes in biofilm cells. These differences suggested that there would be different functional features between biofilms and planktonic cells.
Gene Ontology (GO) analysis of the top 30 most abundant genes in biofilm and planktonic cells revealed that the transcripts associated with biological process of response to cellular protein metabolism were more enriched in planktonic cells than in biofilms (9 genes for biofilm, 22 genes for planktonic). Similarly, the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment revealed that genes involved in ribosomal protein were more enriched in planktonic cells than biofilms (4 genes for biofilm, and 19 genes for planktonic).
Enrichment analysis of differentially expressed genes
Sample to sample distance was calculated using R package DESeq2 (Love and Huber et al. 2014). The principal components analysis (PCA) showed that planktonic and biofilm samples formed distinct clusters (Fig. S2). Meanwhile, the PCA plot displays larger differences among biofilm samples than planktonic samples, suggesting the functional features in the planktonic cells are more conserved than in the biofilm cells. In addition, the differentially expressed genes between biofilm and planktonic cells were also identified using R package DESeq2 Compared to planktonic cells, a total of 1205 (21.63%) genes were identified as differentially expressed in biofilms with 711 (12.76%) genes elevated. Among these genes, the most up-regulated genes were those associated with alkaline phosphatase, the type II hxc secretion system, phosphate ABC transporter, and others. Moreover, 62 of up-regulated genes were not expressed in planktonic samples, including genes associated with the type II hxc secretion system, assimilatory nitrate and nitrite reductases to ammonia, and hypothetical proteins. However, their expression levels in biofilms were not very high, ranged from 1.59 to 451.3 after normalization (Table S5). Therefore, Therefore, these genes could not be suitable to differ biofilms from planktonic cells, but the genes controlled or regulated by these genes could be used as potential biomarker, for example, the expression levels of lapA controlled by the type II hxc secretion system was very high (6237 in biofilm but 2.6 in planktonic).In contrast, transcript levels of 494 (8.87%) genes were down-regulated in biofilms compared to planktonic cells, which involved in denitrification, pyochelin biosynthesis, the type VI secretion system, and others. Of which, 9 genes, related to the type III secretion system, cationic antimicrobial peptide (CAMP) resistance and pyochelin biosynthesis, were not expressed in biofilm samples, but their expression levels in planktonic samples were very low, except that the gene encoding unknown functional protein (Table S6). Therefore, they cannot be used to differ biofilms from planktonic cells.
GO analysis and KEGG pathway enrichment of differentially expressed genes were performed using STRING (Szklarczyk et al. 2015). GO analysis showed that all of three processes were enriched among the genes with decreased transcription in biofilm cells. For biological process, 140 genes were enriched in biological process pathway, followed 132 genes in metabolic process pathway, 109 genes in cellular metabolic process pathway, and 107 genes in organic substance metabolic process pathway. For molecular function process, 136 genes were enriched in molecular function pathway, followed 101 genes in catalytic activity pathway. For cellular component process, 106 genes were enriched in cellular component pathway, and 104 genes in cell pathway. However, no process was enriched among the genes with increased transcription in biofilms, suggesting that biofilms were less metabolically active than planktonic cells. The KEGG pathways enrichment revealed that phosphate and phosphinate metabolism pathway and bacterial chemotaxis pathway were enriched among the transcripts that were elevated in biofilm cells; while two-component system pathway, ribosome, oxidative phosphorylation, and nitrogen metabolism were enriched among down-regulated genes. Interestingly, the processes and the pathways enriched among up-regulated genes were totally different from those among down-regulated genes, which indicate that there would be different metabolism between biofilm and planktonic cells.
Among up-regulated genes, three clusters were found, including one cluster related to the type II hxc secretion system comprising of 22 nodes with 53 edges, one associated with oxidative phosphorylation comprising of 13 nodes with 49 edges, and one involved in bacterial chemotaxis comprising of 8 nodes with 27 edges (Fig. S3). Among down-regulated genes, three clusters were also found, including a cluster involved in denitrification comprising of 20 nodes with 70 edges, one involved in pyochelin biosynthesis comprising of 11 nodes with 24 edges, and one involved in the type VI secretion system comprising of 13 nodes with 35 edges (Fig. S4).
The top 20 most significantly up- and down-regulated genes in biofilms
The top 20 differentially expressed genes in both conditions, with the expression levels of these genes in either condition greater than 100 after normalization, are displayed in Table 1, and ranked according to the log2FC values. Interestingly, among the transcripts elevated in biofilms, 13 genes were clustered with medium confidence (0.40), including 8 genes involved in phosphonate and phosphinate metabolism pathway with highest confidence (0.90), alkaline phosphatase and pyrophosphate specific outer membrane porin OprO (Fig. S5). Of which, lapA was the most up-regulated (log2FC = 11.32) with one of the top 30 most highly expressed genes. Another 7 genes without connected nodes were annotated hypothetical proteins.
Table 1
The top 20 up- and down-regulated genes in biofilm samples, respectively.
Gene_id
|
Gene
|
Log2FC
(PAO_B/PAO_P)
|
PAO_B
|
PAO_P
|
Gene description
|
PA0688
|
lapA
|
11.32
|
6237.32
|
2.60
|
alkaline phosphatase L
|
PA3219
|
PA3219
|
9.85
|
395.65
|
0.46
|
hypothetical protein
|
PA0842
|
PA0842
|
9.60
|
451.26
|
0.62
|
glycosyl transferase family protein
|
PA3383
|
PA3383
|
9.18
|
767.95
|
1.42
|
phosphonate ABC transporter substrate-binding protein
|
PA0691
|
PA0691
|
8.94
|
203.30
|
0.45
|
hypothetical protein
|
PA4382
|
PA4382
|
8.72
|
442.90
|
1.13
|
hypothetical protein
|
PA3382
|
phnE
|
8.71
|
267.00
|
0.69
|
phosphonate transporter PhnE
|
PA4350
|
PA4350
|
8.65
|
1004.08
|
2.68
|
hypothetical protein
|
PA3380
|
PA3380
|
8.29
|
243.00
|
0.83
|
hypothetical protein
|
PA4623
|
PA4623
|
8.17
|
2857.47
|
10.66
|
hypothetical protein
|
PA3378
|
PA3378
|
8.11
|
284.24
|
1.09
|
hypothetical protein
|
PA3381
|
PA3381
|
8.09
|
203.83
|
0.80
|
transcriptional regulator
|
PA3376
|
PA3376
|
8.01
|
188.62
|
0.79
|
phosphonate C-P lyase system protein PhnK
|
PA3280
|
oprO
|
8.01
|
1907.03
|
7.95
|
pyrophosphate-specific outer membrane porin OprO
|
PA3377
|
PA3377
|
8.00
|
374.09
|
1.57
|
hypothetical protein
|
PA3384
|
phnC
|
7.99
|
406.79
|
1.71
|
phosphonate ABC transporter ATP-binding protein
|
PA0692
|
PA0692
|
7.97
|
218.80
|
0.93
|
hypothetical protein
|
PA4354
|
PA4354
|
7.96
|
8259.51
|
36.35
|
hypothetical protein
|
PA0694
|
exbD2
|
7.96
|
260.44
|
1.14
|
transporter ExbD
|
PA2803
|
PA2803
|
7.95
|
159.82
|
0.70
|
hypothetical protein
|
PA1555.1
|
ccoQ2
|
-4.94
|
72.06
|
2569.51
|
cytochrome C oxidase cbb3-type subunit CcoQ
|
PA5171
|
arcA
|
-4.96
|
127.12
|
4229.36
|
arginine deiminase
|
PA5373
|
betB
|
-4.99
|
27.43
|
927.48
|
betaine aldehyde dehydrogenase
|
PA5172
|
arcB
|
-5.00
|
94.84
|
3244.08
|
ornithine carbamoyltransferase
|
PA0516
|
nirF
|
-5.01
|
31.55
|
1090.83
|
heme d1 biosynthesis protein NirF
|
PA4221
|
fptA
|
-5.10
|
4.54
|
165.71
|
Fe(III)-pyochelin outer membrane receptor
|
PA0518
|
nirM
|
-5.11
|
39.30
|
1493.16
|
cytochrome C-551
|
PA0049
|
PA0049
|
-5.19
|
3.95
|
153.21
|
hypothetical protein
|
PA5173
|
arcC
|
-5.31
|
73.26
|
3107.41
|
carbamate kinase
|
PA0519
|
nirS
|
-5.41
|
60.36
|
2744.40
|
nitrite reductase
|
PA2109
|
PA2109
|
-5.86
|
1.75
|
107.96
|
hypothetical protein
|
PA5372
|
betA
|
-5.99
|
3.10
|
209.46
|
choline dehydrogenase
|
PA4225
|
pchF
|
-6.37
|
1.61
|
141.55
|
pyochelin synthetase
|
PA4230
|
pchB
|
-6.48
|
1.44
|
144.11
|
isochorismate-pyruvate lyase
|
PA4888
|
desB
|
-6.94
|
1.23
|
160.81
|
acyl-CoA desaturase
|
PA2114
|
PA2114
|
-7.02
|
4.57
|
631.58
|
major facilitator superfamily transporter
|
PA2111
|
PA2111
|
-7.26
|
5.09
|
835.52
|
hypothetical protein
|
PA4889
|
PA4889
|
-7.30
|
1.52
|
256.77
|
oxidoreductase
|
PA2113
|
opdO
|
-7.94
|
1.88
|
492.54
|
pyroglutatmate porin OpdO
|
PA2112
|
PA2112
|
-7.96
|
2.10
|
561.56
|
hypothetical protein
|
Among the transcripts down-regulated in biofilm, the expression level of nirS gene, important for denitrification pathway, as well as the genes (nirF, nirM) with the functions of electron transport chain for denitrification pathway, was sharply decreased (log2FC < -5.0). Therefore, NO, the product of nitrite reductase NirS, could be important to distinguish biofilms from planktonic cells. In addition, three genes associated with arginine deiminase pathway were also sharply down-regulated (log2FC about -5.00) suggesting that biofilms had significantly decreased level of metabolism.
To identify our RNA-seq analysis, we verified parts of these up- and down-regulated genes using RT-qPCR experiment. Expression levels of these tested genes were consistent with the data obtained from RNA-seq experiment except the expression of pstS (Fig. S6). RNA-seq data analysis revealed that no single gene was found as very high expression in one form of bacterial organization but completely absent in another one. However, in our data, the expression levels of those genes associated with the type II hxc secretion system, phosphate metabolism, and denitrification were significantly changed in biofilm state compared to planktonic state. Therefore, transcription levels of parts of these genes related to above functions were further confirmed with the RT-qPCR experiments on biofilms incubated in the porcine skin explant wells for 24 h, 48 h, and 72 h, respectively. The results demonstrated that expression levels of lapA were higher at the whole biofilm developing cycle than at planktonic state; while the transcription levels of nirS were decreased compared to planktonic state (Fig. 4). Incredibly, the expression of pstS was decreased in biofilms compared to planktonic state. Therefore, alkaline phosphatase LapA would serve as potential marker to monitor chronic wound infections by P. aeruginosa biofilms, and inducing NO or nitrite reductase would be used to inhibit chronic wound infections due to P. aeruginosa biofilms.
Alkaline phosphatase is required for biofilm formation
We also analyzed the extracellular alkaline phosphatase activity that was present in the supernatants of biofilms. Interestingly, the amount of secreted alkaline phosphatase was highest in mature biofilm, following in dispersal biofilms, and very less in early biofilms. However, PAO1 did not show significant alkaline phosphatase production in the supernatant at planktonic state (Fig. 5). In addition, alkaline phosphatase was not found in porcine skin explants as negative controls. Therefore, the results indicated that the release of alkaline phosphatase could contribute to P. aeruginosa biofilm formation in chronic wounds.
Intracellular NO level is reduced in biofilms
Both RNA-seq and RT-qPCR results showed that the transcription of nirS was decreased in biofilms compared to at planktonic state. Therefore, intracellular NO level was detected. The results showed that intracellular NO level was very high when PAO1 was at planktonic state. However, as biofilms growing, the intracellular NO level was going down. After biofilms were in maturation, the NO level was very low; while the NO level was a bit increased in dispersal biofilms (Fig. 6). Therefore, inducing intracellular NO production could be used to inhibit P. aeruginosa biofilms formation in chronic wounds.