Pathogenic diversification of the gut commensal Providencia alcalifaciens via acquisition of a second type III secretion system

ABSTRACT

Providencia alcalifaciens is a Gram-negative bacterium found in various water and land environments and organisms, including insects and mammals. Some P. alcalifaciens strains encode gene homologs of virulence factors found in pathogenic Enterobacterales members, such as Salmonella enterica serovar Typhimurium and Shigella flexneri. Whether these genes are pathogenic determinants in P. alcalifaciens is not known. In this study, we investigated P. alcalifaciens–host interactions at the cellular level, focusing on the role of two type III secretion systems (T3SS) belonging to the Inv-Mxi/Spa family. T3SS_1b is widespread in Providencia spp. and encoded on the chromosome. A large plasmid that is present in a subset of P. alcalifaciens strains, primarily isolated from diarrheal patients, encodes for T3SS_1a. We show that P. alcalifaciens 205/92 is internalized into eukaryotic cells, lyses its internalization vacuole, and proliferates in the cytosol. This triggers caspase-4-dependent inflammasome responses in gut epithelial cells. The requirement for the T3SS_1a in entry, vacuole lysis, and cytosolic proliferation is host cell type-specific, playing a more prominent role in intestinal epithelial cells than in macrophages or insect cells. In a bovine ligated intestinal loop model, P. alcalifaciens colonizes the intestinal mucosa and induces mild epithelial damage with negligible fluid accumulation in a T3SS_1a- and T3SS_1b-independent manner. However, T3SS_1b was required for the rapid killing of Drosophila melanogaster. We propose that the acquisition of two T3SS has allowed P. alcalifaciens to diversify its host range, from a highly virulent pathogen of insects to an opportunistic gastrointestinal pathogen of animals.

INTRODUCTION

Despite belonging to a large and clinically significant family of Gram-negative bacteria, Providencia species remain among the least studied members of Enterobacteriaceae. Providencia spp. colonize diverse hosts and environments. In addition to being found in soil (1), water (2), sewage (3), and retail meats, fruits, and vegetables (4–7), Providencia spp. are members of the human gut, oral cavity, and sputum microbiomes (8–12). Providencia spp. have also been isolated from numerous animals, including penguins, turtles, sharks, snakes (13), nematodes (14), and insects such as blow flies, fruit flies, house flies, and olive flies (13, 15–18). Notably, some Providencia spp. are pathogenic to Drosophila melanogaster, Ceratitis capitata (Mediterranean fruit fly), and Anastrepha ludens (Mexican fruit fly), with the most highly virulent species being Providencia alcalifaciens and Providencia sneebia in D. melanogaster (13) and P. alcalifaciens and Providencia rustigianii in A. ludens (19).

Considered opportunistic bacterial pathogens of humans, P. alcalifaciens, Providencia rettgeri, and Providencia stuartii are the most common clinical isolates (20, 21) and cause a spectrum of nosocomial and environmentally acquired diseases, including urinary tract, wound and ocular infections, diarrhea, meningitis, and sepsis. Providencia spp. are typically resistant to penicillins, first-generation cephalosporins, aminoglycosides, tetracyclines, and polymyxins (22, 23). Increasing antimicrobial resistance is a major public health concern (24, 25). Several groups have reported P. alcalifaciens to be a cause of diarrhea in infants and travelers in developing countries (21, 26–30) and in foodborne-associated outbreaks (31–33). The incidence of P. alcalifaciens in diarrheal patients in Thailand (1.9%), Bangladesh (2.1%) and Kenya (3.2%) is on par with Salmonella spp. (4, 29, 34). A higher incidence of P. alcalifaciens (10%–18%) has been reported for persons with traveler’s diarrhea (21, 35). P. alcalifaciens has also been associated with diarrhea in dogs and cats (36–39). Previous work has verified the ability of some P. alcalifaciens clinical isolates to elicit diarrheal disease in a removable intestinal tie adult rabbit diarrhea (RITARD) infection model (40, 41) and cause fluid accumulation in rabbit ileal loops (31) and diarrhea in suckling mice (42). Furthermore, clinical strains isolated from patients with diarrhea exhibit varying invasive abilities, with some P. alcalifaciens being highly invasive for human epithelial cell lines e.g., HeLa, HEp-2, Vero and Caco-2, and others being noninvasive (31, 43–49). Despite a strong association with diarrheal illness in humans and animals, a detailed understanding of the pathogenic mechanisms of P. alcalifaciens is lacking.

Enteric pathogens such as Salmonella enterica serovar Typhimurium (S. Typhimurium) and Shigella flexneri use type III secretion systems (T3SSs), also known as injectisomes, to deliver “effector” proteins that modulate the actin cytoskeleton, allowing for efficient bacterial entry into non-phagocytic cells. T3SSs are found in many (but not all) Providencia isolates, including P. alcalifaciens (50), an indication of the pathogenic potential of members of this genus. We reported earlier that P. alcalifaciens 205/92, a clinical isolate, encodes for two T3SSs belonging to the Inv-Mxi/Spa family, which we designated T3SS_1a and T3SS_1b (51). Sodalis glossinidius (52), an insect endosymbiont, and some isolates of Providencia spp. are the only other bacteria known to encode for two Inv-Mxi/Spa T3SS. P. alcalifaciens T3SS_1a is closely related to, and functionally interchangeable with, the invasion-associated T3SS1 from S. Typhimurium (51). Structural proteins of T3SS_1b share significant amino acid sequence identity to those of the Ysa T3SS from Yersinia enterocolitica (51), which is restricted to biotype 1B (53). Neither Ysa nor T3SS_1b translocator operons functionally substitute for those of S. Typhimurium in driving bacterial entry into non-phagocytic cells, suggesting an evolutionary functional divergence within the Inv-Mxi/Spa family of T3SSs (51).

We hypothesized that T3SSs are virulence determinants in P. alcalifaciens. In this study, we have investigated the role of the two T3SSs in bacterial colonization of mammalian and insect cells, specifically human intestinal epithelial cells (IECs), human macrophages, and D. melanogaster macrophage-like cells. We report that P. alcalifaciens 205/92 enters eukaryotic cells, rapidly lyses its internalization vacuole, and then replicates within the cytosol. We further show that T3SS_1a is encoded on a 128-kb plasmid and is necessary for efficient bacterial entry, nascent vacuole lysis, and intracellular replication in human IECs. In human macrophages, a T3SS_1a mutant induces less host cell cytotoxicity than wild-type bacteria. While a T3SS_1b mutant exhibits no colonization defect in mammalian or insect cell lines, it is significantly attenuated for the infection of D. melanogaster. Therefore, P. alcalifaciens 205/92 uses two type III injectisomes to colonize diverse eukaryotic hosts.

RESULTS

P. alcalifaciens 205/92 genome

Seven P. alcalifaciens genomes have been sequenced as part of the NIH Common Fund Human Microbiome Project (54) and deposited in GenBank, including P. alcalifaciens 205/92 (https://www.ncbi.nlm.nih.gov/datasets/genome/GCF_000527335.1/). This strain was initially isolated from a young Bangladeshi boy having diarrhea (43, 46). The contig-level genome assembly includes 88 contigs. We generated a complete P. alcalifaciens 205/92 genome using a combination of Oxford Nanopore Technologies (ONT) long-read and Illumina short-read sequencing. Assembly of the highest-quality 100 x filtered and trimmed Nanopore reads (see Methods section) using Trycycler generated three circular replicons: a 4,094,134-bp chromosome and two plasmids, 127,796 bp (p128kb) and 40,541 bp (p41kb) in size. Due to size selection, smaller plasmids are often under-represented in long-read-only assemblies. To account for this, we performed an independent hybrid assembly of the Nanopore reads with bbduk-trimmed Illumina reads, which allowed us to recover a third plasmid, 3,997 bp in size (p4kb). Polishing the resulting combined assembly with Polypolish using the bbduk-trimmed Illumina reads corrected three errors in the chromosome and two in the 128-kb plasmid. The GenBank accession number is GCA_038449115.1.

The P. alcalifaciens 205/92 complete genome has a G + C content of 41.8%, in line with the average G + C content of Providencia spp. genomes (55). Automated annotation of the assembled genome using Bakta identified 4,007 protein-coding genes (3,842, 113, 46, and six proteins coded on the chromosome, p128kb, p41kb, and p4kb plasmids, respectively), as well as seven ribosomal operons (22 rRNA genes overall), and 80 tRNA genes. Using the PHASTER prophage-prediction web tool followed by manual curation, we identified nine putative prophage regions (Fig. 1A). Genomic islands encoding flagella, a type VI secretion system (T6SS) and a type III secretion system, T3SS_1b, were present on the chromosome (Fig. 1A). Genomic islands associated with a type IV secretion system (T4SS) and a second type III secretion system, T3SS_1a (Fig. 1B), were found on p41kb and p128kb, respectively.

Fig 1.

Providencia alcalifaciens 205/92 genome. (A) BRIG diagram of chromosome alignment of seven P. alcalifaciens strains (2019–01-3283-1-1, wls1940, FDAARGOS_408, 2019–04-29292-1-3, PAL-1, 2019–04-29291-1-1, and LHC2-1) and P. rustigianii strain 52579_F01. The complete genome of strain 205/92 presented in this work is used as a reference. Shades of red indicate the phylogenetic branch that includes 205/92; shades of blue indicate the second major branch. Exact coordinates of putative prophages P1–P9, flagella, and secretion systems are provided in Table S1. (B) BRIG diagram of p128kb alignment from nine P. alcalifaciens strains (2939/90, wls1924, wls1925, 2019–04-29292-1-3, 2019–04-29290-1-7, PAL-2, RIMD 1656011, PAL-1, and GCA_902375285.1). p128kb of the 205/92 strain presented in this work is used as a reference. Shades of blue indicate strains from the second phylogenetic branch that does not include 205/92. Exact coordinates of T3SS_1a are provided in Table S1. (C) Whole-genome phylogenetic tree generated with IQTree from the core genome alignment of 46 P. alcalifaciens strains and P. rustigianii strain 52579_F01 used as an outgroup. The bold font indicates complete genome assemblies; the italic font indicates genomes assembled into multiple contigs or scaffolds. The heatmap indicates the presence (complete or partial) or absence of the three plasmids present in the P. alcalifaciens 205/92 genome. The plasmid is classified as complete if the read coverage is >80%; as partial if the read coverage is 40%–80%; and absent if read coverage is <40%. Red asterisks denote isolates collected from human or canine patients with diarrhea (see Table S1 for detailed information on strains). Note that source information is available for all strains. (D) A heatmap of pairwise core SNP distances between the 45 P. alcalifaciens strains and P. rustigianii strain 52579_F01. The total number of identified high-quality (AGCT-only) variable positions is 44,562.

We compared the P. alcalifaciens 205/92 genome with other available P. alcalifaciens genomes (Fig. 1). Core genome alignment identified 44,562 high-quality (AGCT-only) variant positions alongside 3,937,741 constant sites. Phylogenetic analysis of the resulting whole-genome alignment using IQTree (www.iqtree.org/doc/Substitution-Models) identified an TVM + F + I + R2 model as best fitting the data according to the Bayesian information criterion. P. rustigianii strain 52579_F01 was used as an outgroup. Phylogenetic analysis of the strains (Fig. 1C) generally confirmed the conclusions made previously based on protein-coding-based phylogeny (50): strains designated as P. alcalifaciens could be divided into two highly distinct clusters (Fig. 1C and D). Cluster 1, which includes strain 205/92 and a closely related strain 2939/90 also isolated in Bangladesh from a child with diarrhea (40), has much closer and fewer differences between the sequenced genomes and far more isolated strains than cluster 2, which could be the result of a fast and recent spread of Cluster 1. We checked for the presence of T3SS_1a and T3SS_1b in different P. alcalifaciens strains that were chosen according to the phylogeny from the two major clusters. The chromosomal-encoded T3SS_1b was fully present in all the profiled strains (Fig. 1A). For the plasmid-encoded T3SS_1a, our analysis was limited to those strains with the largest plasmid (Fig. 1B and C). Cluster 1 is characterized by a much higher presence of the two large plasmids identified in the 205/92 strain. Indeed, out of 34 isolates that belong to cluster 1, four carried a full (defined as read coverage is >80%), and 13 carried a partial representation of the 128-kb plasmid (read coverage is 40%–80%). Notably, there was a strong correlation between isolates from diarrheal patients, either humans or dogs (indicated by red asterisks in Fig. 1C), and the presence of p128kb (Fig. 1C). These isolates were sourced from diverse geographical locations (Table S1). In contrast, only three out of 11 isolates from cluster 2 carried a partial p128kb. We were surprised to find that, despite the considerable differences in the overall genetic content of this plasmid between strains, the coding potential for all T3SS_1a structural and regulatory proteins remained intact in all the strains (Fig. 1B; Fig. S1). Interestingly, there was considerable heterogeneity between strains in the protein-coding sequence length for a predicted type III effector, SipA. In P. alcalifaciens 205/92, SipA is a > 2,300 AA protein (Fig. 1B). Nine out of 34 isolates from cluster 1 and none from cluster 2 carried the 41-kb plasmid. The small p4kb plasmid had a more uniform distribution: 5/34 for cluster 1 and 3/11 for cluster 2 (Fig. 1C). A mean distance between isolates from cluster 1 and cluster 2 was determined to be between 16,000 and 17,000 SNPs per 44,562 high-quality variable positions. Cluster 1 had a mean distance between the two strains of approximately 1,000 SNPs; at the same time, the mean distance between the two strains in cluster 2 was over 5,000 SNPs (Fig. 1D). Overall, we conclude that the P. alcalifaciens species is genomically diverse and consists of two major lineages (50).

T3SS_1a genes are induced in the late-log phase of growth

Given the involvement of secretion systems in the pathogenesis of numerous Gram-negative bacteria, we initially set out to define in vitro growth conditions under which P. alcalifaciens T3SS genes are transcriptionally active. The upstream regulatory regions of invF, prgH, and sicA, the first genes in predicted operons from T3SS_1a and T3SS_1b pathogenicity islands (Fig. 2A), were cloned upstream of a promoterless luxCDABE in pFU35. The resulting plasmids were electroporated into wild-type (WT) P. alcalifaciens. We detected robust luminescence for bacteria carrying PprgH_1a-luxCDABE and PinvF_1a-luxCDABE transcriptional reporters at the late log-phase of growth in LB–Miller broth, pH 7.0, at 37°C (Fig. 2B and C). Transcriptional activity for sicA_1a was much lower, but still greater than that of the promoter-less vector, pFU35, under these conditions (Fig. 2C). By contrast, no luminescence was detected for bacteria carrying T3SS_1b gene reporters under any of the in vitro growth conditions we tested, i.e., shaking in LB–Miller broth, pH 7.0 (Fig. 2C) or pH 5.8 at 37°C or 25°C; shaking in M9 minimal media pH 7.0 at 37°C or 25°C; or McCoy’s medium or Schneider’s medium in the absence or presence of 10% heat-inactivated calf serum at 25°C and 37°C (data not shown). Collectively, we conclude that T3SS_1a- and T3SS_1b-associated genes are expressed under distinct conditions.

Fig 2.

Characterization of T3SS_1a and T3SS_1b in P. alcalifaciens 205/92. (A) Cartoon depiction of the genetic organization of T3SS_1a and T3SS_1b. (B) Growth curve of P. alcalifaciens 205/92 WT and WT carrying pFU35 plasmid. Bacterial subcultures were grown at 37°C, with shaking at 220 rpm in LB–Miller broth, and OD₆₀₀ measured hourly. Mean ± SD of three independent experiments. (C) Bacterial luminescence over a time course of growth. P. alcalifaciens 205/92 carrying luxCDABE transcriptional reporters for the indicated T3SS_1a (upper panel) or T3SS_1b (lower panel) gene promoters or the empty plasmid (pFU35) were grown as in (B), and relative light units (RLU) were measured every hour by using a microplate reader. Lines indicate the mean of three independent experiments (each symbol represents data from one experiment). (D) Secreted protein profile. P. alcalifaciens WT, ∆invA_1a, and ∆invA_1b subcultures were grown for 4 hours in LB–Miller broth. Supernatants were collected, filtered, and precipitated proteins separated by SDS-PAGE and stained with GelCode Blue. Molecular mass markers are shown on the left. The protein bands indicated by asterisks were excised and identified as SipA_1a (>250 kDa), SipC_1a (~40 kDa), and SipD_1a (~40 kDa) by mass spectrometry.

Salmonella spp. secrete effector proteins into culture media in a type III secretion-dependent manner (56). When P. alcalifaciens was grown under in vitro conditions that induce T3SS_1a-associated genes (late-log phase in LB–Miller broth), numerous proteins were secreted into the culture supernatant (Fig. 2D). To identify whether protein secretion was dependent on T3SS_1a or T3SS_1b, we compared the protein profiles of culture supernatants from WT, ∆invA_1a, and ∆invA_1b bacteria. InvA is a highly conserved inner membrane component of the Inv-Mxi/Spa T3SS family. A S. Typhimurium invA mutant is unable to assemble the needle portion of the injectisome and, as a result, is deficient for T3SS-dependent protein secretion (57). By analogy, we presume that P. alcalifaciens invA deletion mutants are type III secretion-incompetent. The secreted protein profile of ∆invA_1b bacteria was indistinguishable from that of WT bacteria. In contrast, two major protein bands, one at >250 kDa and one at ~40 kDa, and numerous minor protein bands were absent from the supernatants of ∆invA_1a bacteria (Fig. 2D). Mass spectrometric analysis identified the most abundant proteins as SipA_1a (predicted molecular mass of 240 kDa) and a mixture of SipC_1a and SipD_1a (predicted mass of 43 and 39 kDa, respectively), the corresponding genes of which are encoded on T3SS_1a (Fig. 2A). In Salmonella spp., SipA is a type III effector with actin-binding properties (58), SipC is a type III translocator protein (59), and SipD is the needle tip protein of T3SS1 (59). Due to the presence of tandem repeat sequences, P. alcalifaciens 205/92 SipA_1a is much larger than orthologous proteins from S. Typhimurium (SipA), Shigella flexneri (IpaA) (~74 kDa), and some other P. alcalifaciens strains. Of the 20 P. alcalifaciens strains harboring p128kb, 11 encode for SipA_1a of >200 kDa (Fig. 1B; Fig. S1).

Providencia spp. have peritrichous flagella and are considered motile. P. alcalifaciens 205/92 encodes numerous flagella-associated genes in one large genetic island on the chromosome (Fig. 1A; Fig. S2A). We constructed luxCDABE-based transcriptional reporters to the upstream regulatory regions of flhD, flgB, and fliC and measured bacterial luminescence when WT bacteria carrying these reporters were grown in LB–Miller broth pH 7.0 at 37°C for 12 hours. All three genes were transcribed, with a peak of transcription at the late log-phase of growth (Fig. S2B). Swimming motility of P. alcalifaciens was evident on soft agar plates, albeit much less than that of S. Typhimurium WT bacteria but greater than that of a non-motile S. Typhimurium ∆flgB mutant (Fig. S2C). Overall, we conclude that P. alcalifaciens T3SS_1a and flagellar genes are induced by aeration and at the late log-phase of growth, similar to T3SS1 and flagellar genes in S. Typhimurium (60).

Internalization of P. alcalifaciens into mammalian and insect cell lines

Some clinical isolates of P. alcalifaciens, including 205/92, have been reported to invade HEp-2 cell monolayers (43, 46). To investigate the phenotypic characteristics of the initial interaction of P. alcalifaciens 205/92 with non-phagocytic mammalian cells, we used scanning electron microscopy (SEM) and transmission electron microscopy (TEM). P. alcalifaciens were grown under conditions that induce the T3SS_1a (Fig. 2C), added to monolayers (HeLa or HCT116 epithelial cells), centrifuged for 5 minutes, and incubated for a further 15 minutes (20 minutes post-infection (p.i.) for SEM) or 55 minutes (1 h p.i. for TEM) at 37°C, and then infected monolayers were processed for microscopy. By SEM, we observed P. alcalifaciens attaching to filopodial extensions on the epithelial cell surface (Fig. 3A and C). Sometimes, these membrane protrusions were wrapped around the bacteria (Fig. 3A). Similar initial interactions with epithelial cells have been described for S. flexneri, Y. enterocolitica, and Helicobacter pylori (61–64). Invasion of P. alcalifaciens into epithelial cells is associated with actin condensation at the site of bacterial entry (40) and inhibited by cytochalasin D (43), a hallmark of the “trigger” type of cell entry mediated by S. Typhimurium and S. flexneri. By SEM, we did not observe dramatic plasma membrane ruffles, which are characteristic of bacterial entry via this mechanism, however (Fig. 3C). TEM analysis suggested that P. alcalifaciens internalization into non-phagocytic cells was instead via a zipper-like mechanism; upon bacterial adherence, membrane protrusions formed and wrapped around bacteria in tight apposition, eventually engulfing the entire bacterium (Fig. 3B and D).

Fig 3.

P. alcalifaciens adherence to, and invasion of, eukaryotic cells. (A) Scanning electron micrograph (SEM) of P. alcalfaciens 205/92 associating with filopodia on the surface of a HeLa epithelial cell at 20 minutes p.i. The scale bar is 1 µm. (B) Transmission electron micrograph (TEM) showing P. alcalifaciens entering a HeLa cell at 1 h p.i. The scale bar is 0.5 µm. (C) SEM shows bacteria associating with filopodia on the surface of a HCT116 colonic epithelial cell at 20 minutes p.i. One bacterium is seen entering via a zipper-like mechanism (indicated by arrowhead). The scale bar is 1 µm. (D) TEM showing P. alcalfaciens entering a HCT116 cell at 1 h p.i. The scale bar is 0.5 µm. (E) HCT116, THP-1, and S2 cells were infected with bacterial subcultures at MOIs of ~150, ~10, and ~5, respectively. Invasion efficiencies of P. alcalifaciens WT, ∆invA_1a, ∆invA_1a carrying pGEN-MCS (empty vector) or pGEN-invA_1a (complemented strain), and ∆invA_1b bacteria were compared by gentamicin protection assay at 1 h p.i. Invasion efficiency is defined as the percentage of the inoculum that was internalized; the invasion efficiency of WT bacteria was normalized to 100% in each experiment. Mean ± SD from three to four independent experiments. Asterisks indicate data significantly different from WT or ∆invA_1a pGEN bacteria (P < 0.05, ANOVA with Dunnett’s post-hoc test).

As P. alcalifaciens is known to colonize mammalian and insect hosts, we tested whether its entry into different types of eukaryotic cells was T3SS-dependent. P. alcalifaciens 205/92 WT and ∆invA deletion mutants were grown under conditions that induce the T3SS_1a (Fig. 2C) before their addition to mammalian and insect cells. Internalized bacteria were enumerated at 1 h p.i. using the gentamicin protection assay, and invasion efficiency was calculated as the proportion of the bacterial inoculum that was internalized. Compared to WT bacteria (set to 100%), ∆invA_1a bacteria were highly defective for entry into HCT116 human colonic epithelial cells (15 ± 5.7%; Fig. 3E), a phenotype that could be partially restored by plasmid-borne complementation (Fig. 3E). Deletion of invA_1a also affected P. alcalifaciens entry into D. melanogaster S2 cells (31 ± 9.6%; Fig. 3E), which have characteristics of fly hemocytes. In macrophages derived from the human monocytic cell line, THP-1, there was no difference in the internalization efficiency of the three bacterial strains (Fig. 3E) in accordance with the phagocytic properties of these cells. By contrast, the invasion efficiency of ∆invA_1b bacteria was equivalent to that of WT bacteria in HCT116, THP-1, and S2 cells (Fig. 3E). Overall, we conclude that P. alcalifaciens T3SS_1a is required for efficient bacterial internalization into human IECs and insect cells, suggesting a role for the T3SS_1a-dependent translocation of type III effectors in this entry process.

Intracellular replication of P. alcalifaciens

We next investigated whether P. alcalifaciens 205/92 can survive and replicate intracellularly in mammalian and insect cells. In HCT116 epithelial cells, there was a 3.3-fold increase in recoverable CFUs for WT bacteria over a 12-h time course (Fig. 4A, left panel), with the greatest net increase in intracellular proliferation occurring between 1 h p.i. and 4 h p.i. (Fig. 4A, left panel). We confirmed the intracellular proliferation of WT bacteria within epithelial cells using an inside/outside assay in conjunction with fluorescence microscopy. HCT116 cells were infected with P. alcalifaciens carrying a plasmid constitutively expressing dsRed (pGEN-DsRed.T3), and anti-P. alcalifaciens antibodies were used to probe for extracellular bacteria in non-permeabilized cells. The number of intracellular WT bacteria increased from 1 h p.i. (mean of 1.7 bacteria/cell) to 8 h p.i. (mean of 2.92 bacteria/cell) (Fig. 4A, middle panel). Replication of the ∆invA_1b mutant was indistinguishable from that of WT bacteria in HCT116 cells (Fig. 4A , left and middle panels). However, a progressive decrease in recoverable CFUs for ∆invA_1a bacteria was observed in the gentamicin protection assay (Fig. 4A, left panel) and there was no evidence of bacterial proliferation at the single-cell level (mean of 1.6 bacteria/cell and 1.8 bacteria/cell at 1 h p.i. and 8 h p.i., respectively) (Fig. 4A, middle panel). Negligible lactate dehydrogenase (LDH) release, a measure of cytotoxicity, was detected over the infection time course irrespective of the infecting bacterial strain (Fig. 4A, right panel).

Fig 4.

Intracellular replication of P. alcalifaciens in eukaryotic cells. (A) Intracellular proliferation of WT, ∆invA_1a, and ∆invA_1b bacteria in HCT116 cells was quantified by the gentamicin protection assay (left panel, n = 3 independent experiments, where each dot represents the mean from one experiment) and fluorescence microscopy (middle panel; each dot represents the number of bacteria in one cell, and the horizontal bar indicates the mean of three independent experiments). Infections conditions for WT and ∆invA_1b bacteria were as described for Fig. 3E (MOI ~ 150). The MOI for ∆invA_1a bacteria was increased by two- to threefold (MOI ~ 300–450) to increase bacterial internalization. For microscopy experiments, WT, ∆invA_1a, and ∆invA_1b bacteria were carrying pGEN-DsRed.T3. Inside/outside staining was used to distinguish intracellular from extracellular DsRed-labeled bacteria. Cell death was measured by LDH release into the supernatants (right panel, n = 3 independent experiments, where each dot represents the mean from one experiment). Percent cytotoxicity was calculated by normalizing to maximal cell death (1% (vol/vol) Triton X-100 lysis). (B) Intracellular proliferation of WT, ∆invA_1a, and ∆invA_1b bacteria in THP-1 macrophages (MOI ~ 10 for all strains) was quantified by the gentamicin protection assay (left panel, n = 3 independent experiments) and fluorescence microscopy (middle panel, n = 3 independent experiments). Cytotoxicity was measured using LDH release assay (right panel, n = 3 independent experiments). (C) Intracellular proliferation of WT, ∆invA_1a, and ∆invA_1b bacteria in S2 cells (MOI ~ 5 for all strains) was quantified by the gentamicin protection assay (left panel, n = 3 independent experiments) and fluorescence microscopy (middle panel, n = 2 independent experiments).

The patterns of intracellular proliferation for WT bacteria and the deletion mutants were different in human macrophages as compared to human IECs. Using the gentamicin protection assay, there was a decrease in recoverable CFUs for WT bacteria over a 12-h time course of infection in THP-1 cells (Fig. 4B, left panel). Incongruently, there was a slight increase in the mean number of WT bacteria/cell at the single-cell level, from a mean of 1.8 bacteria/cell at 1 h p.i. to 2.6 bacteria/cell at 8 h p.i., (Fig. 4B, middle panel, Fig. S3). However, high levels of cell death/cytotoxicity during WT infection, up to 35% of the monolayer by 12 h p.i. (Fig. 4B, right panel), likely account for the overall decrease in viable CFUs, as assessed by the gentamicin protection assay. ∆invA_1b bacteria showed a similar profile to WT bacteria for CFUs, number of bacteria/cell, and cytotoxicity induction kinetics (Fig. 4B). By contrast, there was a net increase in ∆invA_1a bacteria over time, as measured by CFUs and the number of bacteria per cell (mean of 2.0 and 5.2 bacteria/cell at 1 h p.i. and 8 h p.i., respectively), indicating replication of ∆invA_1a bacteria in THP-1 cells (Fig. 4B). However, significantly less LDH was released into the culture supernatants, suggesting that the overall increase in viable ∆invA_1a bacteria was explained, in part, by decreased THP-1 cell lysis/detachment of infected cells compared to WT infection (Fig. 4B).

Notably, the level of bacterial replication was higher in insect cells than in mammalian cells. Upon P. alcalifaciens infection of D. melanogaster S2 cells, there was an overall increase in intracellular CFUs over 12 hours for WT (6.2-fold), ∆invA_1a (4.2-fold), and ∆invA_1b bacteria (5.6-fold) (Fig. 4C). Similarly, the mean number of bacteria per S2 cell increased from 1 h p.i. to 8 h p.i. for all three strains (2.2 to 4.7 bacteria/cell for WT, 1.5 to 3.8 bacteria/cell for ∆invA_1a, and 2.2 to 3.8 bacteria/cell for ∆invA_1b; Fig. 4C, middle panel, Fig. S3). The colorimetric assay for LDH release cannot be used with S2 cells (65), so we could not assess host cell cytotoxicity concurrently with CFUs for Providencia-S2 cell infections. Collectively, these data indicate that P. alcalifaciens 205/92 can replicate intracellularly in mammalian and insect cells, and the contribution of T3SS_1a to bacterial proliferation and induction of host cell death is both host- and cell-type dependent.

Intracellular expression kinetics of T3SS_1a and T3SS_1b

To follow P. alcalifaciens gene expression after internalization into mammalian and insect cells, we infected HCT116 and S2 cells with WT bacteria carrying plasmid-borne PinvF_1a-luxCDABE or PprgH_1a-luxCDABE as T3SS_1a reporters or PinvF_1b-luxCDABE or PprgH_1b-luxCDABE as T3SS_1b reporters. At various time points p.i., infected monolayers were collected, lysed, and luminescence associated with intracellular bacteria quantified in a plate reader. We found that the intracellular expression kinetics of P. alcalifaciens genes encoded in T3SS_1a are comparable to those in S. Typhimurium T3SS1 (60). Specifically, we observed a rapid downregulation of prgH_1a and invF_1a gene transcription, such that by 6 h p.i., bacterial luminescence was equivalent to that of the promoterless luxCDABE plasmid backbone, pFU35 (Fig. 5). Transcription of prgH_1b and invF_1b was not observed over the time course, indicating that T3SS_1b-associated genes are not induced intracellularly in mammalian or insect tissue culture cells, at least up to 6 h p.i.

Fig 5.

Rapid downregulation of T3SS_1a after bacterial internalization. HCT116 (left panel, MOI ~ 150) or S2 (right panel, MOI ~ 5) cells were infected with subcultures of P. alcalifaciens WT bacteria carrying one of the following transcriptional reporter plasmids: PinvF_1a-luxCDABE, PprgH_1a-luxCDABE, PinvF_1b-luxCDABE, PprgH_1b-luxCDABE, or the empty vector control (pFU35). At the indicated time points, infected monolayers were collected, lysed, and the luminescence associated with internalized bacteria measured in a plate reader. The line indicates the mean of three independent experiments (each symbol represents data from one experiment).

P. alcalifaciens colonizes the cytosol of eukaryotic cells

Our earlier work on the SipC/IpaC family of type III translocators showed that these proteins have differential destabilizing activities on bacteria-containing vacuole membranes (66). For example, replacing SipC in S. Typhimurium with its ortholog from either of the professional cytosolic pathogens, S. flexneri (IpaC) or Chromobacterium violaceum (CipC), enables Salmonella to lyse its internalization vacuole more efficiently. We used this gene swapping strategy to predict the intracellular niche of P. alcalifaciens 205/92. We have previously shown that P. alcalifaciens SipC_1a can partially complement for the internalization defect of a S. Typhimurium ∆sipC mutant (∆sipC::sipC_1a) into non-phagocytic cells (51). Using the chloroquine (CHQ) resistance assay (67, 68), which determines the proportion of internalized bacteria that are present in the cytosol, we found a significantly increased proportion of S. Typhimurium ∆sipC::sipC_1a bacteria in the cytosol of HCT116 epithelial cells and J774A.1 macrophages compared to S. Typhimurium WT (Fig. 6A). Notably, the level of nascent vacuole lysis for ∆sipC::sipC_1a and ∆sipC::cipC bacteria was comparable (Fig. 6A). From these results, we speculated that P. alcalifaciens is a cytosolic bacterium, like C. violaceum (66).

Fig 6.

P. alcalifaciens lyses its internalization vacuole and then replicates in the cytosol of eukaryotic cells. (A) HCT116 (left panel, MOI ~ 50) or J774A.1 (right panel, MOI ~ 5) cells were infected with S. Typhimurium (STm) WT, ∆sipC::cipC (chromosomal replacement of sipC with C. violaceum cipC), or ∆sipC::sipC_1a (chromosomal replacement of sipC with P. alcalifaciens sipC_1a) bacteria, and the proportion of internalized bacteria present in the cytosol at 90 min p.i. was assessed by CHQ resistance assay. Mean ± SD from four independent experiments. Asterisks indicate data significantly different from STm WT bacteria (P < 0.05, ANOVA with Dunnett’s post-hoc test). (B) HCT116 (left panel, MOI ~ 150), THP-1 (middle panel, MOI ~ 10), and S2 (right panel, MOI ~ 5) cells were infected with P. alcalifaciens 205/92 WT, ∆invA_1a, and ∆invA_1b bacteria, and the proportion of cytosolic bacteria at 90 min p.i. and 8 h p.i. was determined by CHQ resistance assay. Mean ± SD from three to four independent experiments. Asterisks indicate data significantly different from WT bacteria (P < 0.05, ANOVA with Dunnett’s post-hoc test). (C and D) Representative TEM images of cytosolic (C) and vacuolar (D) P. alcalifaciens in HeLa epithelial cells at 1 h p.i. Scale bars are 200 nm. (E) Representative TEM image of P. alcalifaciens replicating in the cytosol of a HeLa epithelial cell at 4 h p.i. The scale bar is 500 nm; N = nucleus; arrowheads indicate extracellular bacteria presumably killed by gentamicin.

We applied the CHQ resistance assay to Providencia-infected HCT116, THP-1, and S2 cells to determine their intracellular replication niche unequivocally. At 90 min p.i., ~80% of WT bacteria were present in the cytosol in all three cell types (Fig. 6B). Therefore, lysis of the nascent bacteria-containing vacuole is fast and efficient. This level of cytosolic presence was sustained at later times (>70% cytosolic bacteria at 8 h p.i.; Fig. 6B), indicating bacterial replication (Fig. 4) occurs in the cytosol of eukaryotic cells. There was no significant difference in the proportion of cytosolic WT and ∆invA_1b bacteria in any cell type (Fig. 6B). However, a lower proportion of ∆invA_1a bacteria were present in the cytosol of HCT116 cells at 90 min p.i. and 8 h p.i. (Fig. 6B), indicating a defect in bacteria-containing vacuole lysis in human IECs. This dependence on T3SS_1a for vacuole lysis was not observed in human macrophages or insect cells, however (Fig. 6B).

We used TEM as an independent method to assess the intracellular site of P. alcalifaciens in eukaryotic cells. Most WT bacteria were free in the cytosol of epithelial cells by 1 h p.i. (59% cytosolic in HeLa cells, n = 49, Fig. 6C; 64% cytosolic in HCT116 cells, n = 11). An intact vacuolar membrane was observed around some bacteria in epithelial cells at 1 h p.i. (Fig. 6D). By 4 h p.i., bacterial replication in the cytosol of epithelial cells was evident (Fig. 6E). We also observed bacteria that remain extracellular, seemingly firmly attached to the surface of epithelial cells, which were killed by gentamicin (Fig. 6E). Like epithelial cells, most intracellular WT bacteria were found in the cytosol without any obvious surrounding membrane in THP-1 macrophages (83% cytosolic, n = 18) and S2 cells (93% cytosolic, n = 43) by 90 min p.i. Collectively, these data support that P. alcalifaciens 205/92 rapidly escapes from its internalization vacuole and then replicates in the cytosol of mammalian and insect cells.

Inflammasome activation by P. alcalifaciens

Inflammasomes are cytosolic innate immune sensors that play a key role in restricting bacterial infections, and epithelial-intrinsic inflammasomes mediate protective responses against intestinal pathogens such as S. Typhimurium (69–71) and C. rodentium (72). Considering P. alcalifaciens can invade mammalian cells (Fig. 3 and 4) and replicate in the cytosol (Fig. 6), we speculated it would activate human IEC canonical (caspase-1) and/or non-canonical (caspase-4) inflammasomes. We have previously shown that the activation status of murine IECs affects the contribution of non-canonical and canonical inflammasomes to host defense (73). We therefore considered that inflammasome activation in human IECs might also be similarly affected. In agreement with previous reports in HT-29 colonic epithelial cells (74–76), CASP1 mRNA was significantly upregulated by IFNγ treatment in IECs, specifically HCT116 (Fig. 7A) and Caco-2 C2BBe1 (Fig. S4A) cells. A robust increase in pro-caspase-1 and moderate increase in pro-caspase 4 levels were detected upon IFNγ treatment in HCT116 cells (Fig. 7B). Only pro-caspase-1 was responsive to IFNγ treatment in C2Bbe1 cells (Fig. S4B). Therefore, with IFNγ priming in conjunction with CASP1^-/- and CASP4^-/- knockout (KO) cells, we can investigate whether P. alcalifaciens 205/92 activates canonical and non-canonical inflammasomes in IECs.

Fig 7.

P. alcalifaciens 205/92 activates the non-canonical inflammasome in IECs. (A) HCT116 cells were left untreated or treated with 50 ng/mL IFNγ for 16–18 hours. mRNA expression of CASP1, CASP4, CASP5, IL18, and IL1B relative to the reference gene RPLP0 was measured by qPCR (expressed as 2^-∆Cq). n = three–four independent experiments. The asterisk indicates data significantly different from those of untreated cells, P < 0.05, Student’s t-test. (B) Immunoblot analysis of caspase-1, caspase-4, and ß-actin (loading control) in HCT116 WT, CASP1^-/- (CASP1 KO) or CASP4^-/- (CASP4 KO) cells, left untreated or treated with IFNγ for 16–18 hours. (C) IL-18 (upper panel) and IL-8 (lower panel) release into supernatants from mock-infected or P. alcalifaciens-infected HCT116 WT, CASP1 KO, or CASP4 KO cells (MOI ~150) at 16 h p.i. was quantified by ELISA. n ≥ 8 independent experiments. Cells were treated with IFNγ (50 ng/mL) for 16–18 hours before infection where indicated (+). Asterisks indicate significantly different data, P < 0.05, Student’s t-test. (D) Bacterial replication in HCT116 WT, CASP1 KO, and CASP4 KO cells was assessed by gentamicin protection assay. Fold-replication is CFU_10h/CFU_1h. The asterisk indicates data significantly different from those of HCT116 WT, P < 0.05, ANOVA with Dunnett’s post-hoc test. n = three–four independent experiments.

Processing and secretion of two cytokines, interleukin (IL)−18 and IL-1β, is dependent on inflammasome activation. Despite encoding IL1B mRNA (77) (Fig. 7A), IL-1β is not secreted by human IECs upon bacterial infection ((78); our unpublished results). We therefore used IL-18 secretion as a readout of inflammasome activation. IL-8 release was a control for an inflammasome-independent cytokine. Upon Providencia infection of WT HCT116 cells, IL-18 was detected in the cell culture supernatant at 16 h p.i. in both untreated and IFNγ-primed cells (Fig. 7C). The infection-dependent potentiation of IL-18 release from both untreated and IFNγ-primed cells was statistically significant (Fig. 7C). The magnitude of increase in IL-18 release after P. alcalifaciens infection of CASP1^-/- KO cells was comparable to that of WT HCT116 cells, with and without IFNγ treatment (Fig. 7C). By contrast, IL-18 release was abrogated from infected CASP4^-/- KO cells, irrespective of IFNγ priming (Fig. 7C). Infection-induced IL-8 secretion was comparable regardless of the priming status or cell genotype (Fig. 7C). These data indicate that P. alcalifaciens 205/92 stimulates IL-18 secretion from human IECs in a caspase-4-dependent manner, i.e., via activation of the non-canonical inflammasome.

Previous studies in human IECs have shown that caspase-4 is critical for limiting S. Typhimurium replication (69, 78, 79). To test whether IEC intrinsic inflammasomes (caspase-1 and/or caspase-4) influence intracellular proliferation of P. alcalifaciens 205/92, we compared bacterial replication. In unprimed HCT116 cells, there was no significant difference in P. alcalifaciens replication in WT, CASP1^-/-, or CASP4^-/- cells over 10 hours (9.2-, 7.6-, and 11.9-fold, respectively; Fig. 7D). IFNγ priming efficiently restricted bacterial replication in WT and CASP1^-/- HCT116 cells (3.3- and 3.3-fold, respectively) but not CASP4^-/- cells. Rather, robust bacterial replication was still observed in CASP4^-/- HTC116 cells upon IFNγ-priming (14.5-fold; Fig. 7D). In C2Bbe1 cells, enhanced bacterial proliferation was detected in CASP4^-/- cells, irrespective of IFNγ priming (Fig. S3C). Therefore, caspase-4, but not caspase-1, restricts the intracellular proliferation of P. alcalifaciens in human IECs.

Infection models

We first used the bovine ligated intestinal loop model to assess the enteropathogenicity of P. alcalifaciens 205/92. We have previously used this infection model to study another enteric pathogen, S. Typhimurium (80). Bacteria (~10⁹ CFU) were injected into each loop, and loops were collected at 12 hours post-inoculation. By CFU counts, most bacteria remained extracellular, with only 10⁷ WT bacteria being tissue-associated (Fig. 8A). Compared to the LB control, infection with P. alcalifaciens 205/92 did not promote fluid accumulation in the calf model (Fig. 8B), a proxy for intestinal secretory responses. Infection-induced inflammatory changes were assessed by histological evaluation of hematoxylin- and eosin-stained sections of tissue samples. Of the six criteria scored—polymorphonuclear infiltration to the lamina propria, submucosal edema, epithelial damage, villus blunting, crypt abscess, and cell death (Table 1) —only epithelial damage was increased with statistical significance in infected tissues (P = 0.04, Student’s t-test; Fig. 8C through E). In this infection model, we did not observe any difference in fluid or tissue colonization (Fig. 8F), fluid accumulation or local inflammatory responses (results not shown) when comparing infection with WT, ∆invA_1a, or ∆invA_1b bacteria at 2 h or 8 h p.i.

Fig 8.

P. alcalifaciens 205/92 infection of bovine ligated intestinal loops. Bovine ligated intestinal loops were inoculated with LB broth (control) or P. alcalifaciens 205/92 (~10⁹ CFU) resuspended in LB broth. (A) Recovery of WT bacteria from the fluid and tissue after 12 hours. Each dot represents data from one calf. Means are indicated. (B) Secretory responses in intestinal loops after infection with WT bacteria (P.a.) or LB control (LB) for 12 hours. Each dot represents data from one calf. Means are indicated. (C) Pathological scores from hematoxylin- and eosin-stained loop tissues sampled 12 hours after inoculation with LB broth or WT bacteria. P. alcalifaciens infection led to increased epithelial damage in the intestine (indicated by asterisk, P = 0.040, Student’s t-test). Each vertical bar represents one calf. (D) and (E) Representative images (20 x objective) of hematoxylin- and eosin-stained sections of intestinal tissue from loops inoculated with LB broth alone (D) or WT bacteria (E) for 12 hours. Arrowheads indicate sites of epithelial damage. Scale bars are 50 µm. (F) Recovery of WT, ∆invA_1a, and ∆invA_1b bacteria from the fluid and tissue (cell-associated or intracellular) after 2 hours (left panel) and 8 hours (right panel). Each dot represents the average of two–three loops per calf (2 hours) or one loop per calf (8 hours). Means are indicated.

As an alternative infection model, we used D. melanogaster to test whether T3SS_1a or T3SS_1b were required for virulence. In insect hosts, P. alcalifaciens is highly virulent (13). In previous studies, an infection dose of 10³–10⁴ Dmel, a P. alcalifaciens strain initially isolated from wild-type D. melanogaster, or ~3750 CFU P. alcalifaciens DSM30120, led to fly death in 1–2 days (13, 81). In our initial studies, an infection dose of ~300 CFU for P. alcalifaciens 205/92 proved too high, with rapid killing of all flies between 30 and 40 hours (results not shown). When we reduced the infection dose to 30 CFU, all flies succumbed to infection with WT bacteria within 45–68 hours (Fig. 9). Virulence of P. alcalifaciens ∆invA_1a bacteria was comparable to that of WT bacteria at an infection dose of 30 CFU (Fig. 9, P = 0.83), whereas ∆invA_1b-infected flies showed an extended time to death (Fig. 9, P < 0.0001). P. alcalifaciens-induced mortality was dose-dependent; at a lower infectious dose of WT bacteria (10 CFU), 5%–6% of flies survived for up to 93 hours, when we stopped monitoring survival (Fig. 9). Compared to WT bacteria, we observed a prolonged time to death and decreased mortality during infection with ∆invA_1a and ∆invA_1b bacteria at a dose of 10 CFU (Fig. 9). Both survival curves were significantly different from infection with WT bacteria (P = 0.0019 for ∆invA_1a and P < 0.0001 for ∆invA_1b), indicating that flies are less susceptible to infection with these gene deletion mutants. Overall, these results demonstrate two points: (1) flies are highly susceptible to infection with P. alcalifaciens 205/92 and (ii) T3SS_1a and T3SS_1b are both virulence determinants in an insect host.

Fig 9.

T3SS_1b, and to a lesser extent T3SS_1a, are virulence determinants in insects. D. melanogaster were inoculated with P. alcalifaciens 205/92 WT, ∆invA_1a, or ∆invA_1b bacteria, or PBS, and survival was monitored. Results are from an infectious dose of ~30 CFU/fly (left panel) or ~10 CFU/fly (right panel) and are representative of two–three independent experiments. Asterisks indicate survival curves significantly different from WT infection (P < 0.05, logrank test). >50 flies were inoculated with PBS or each strain in each experiment.

DISCUSSION

T3SS play a central role in cell–cell interactions between bacteria and eukaryotes, irrespective of whether bacteria are pathogens, mutualists, or commensals (82). The high prevalence of T3SS in Providencia spp. suggests this supramolecular system is important for their colonization of specific ecological niches. Up to nine T3SS families have been identified in Gram-negative bacteria (83), and both T3SSs of P. alcalifaciens 205/92 belong to the Inv/Mxi-Spa family. Our earlier phylogenetic analysis of the translocator proteins in this family indicated that there are two sub-groups of T3SS (51). The first sub-group contains bacteria that colonize and cause disease in mammals, i.e., S. enterica, S. flexneri, enteroinvasive E. coli (EIEC), and Chromobacterium spp. P. alcalifaciens T3SS_1a is also in this clade. The second sub-group, with T3SS_1b, contains bacteria that colonize diverse environments, such as pathogens of fungi (Pseudomonas gingeri), plants (Xanthomomas albilineans), insects (Providencia sneebia), endosymbionts of insects (Sodalis glossinidius), and opportunistic pathogens of humans (Proteus mirabilis). Notably, T3SS required for bacterial colonization of insect hosts are restricted to the second sub-group. For example, the SSR-2 T3SS from S. glossinidus enhances bacterial proliferation in insect cells (52), the Ysa T3SS aids in Y. enterocolitica replication in insect, and not mammalian, cells (65), and PSI-2 from Pantoea stewartii is required for persistence in the gut of its flea beetle vector (84). We have further shown that T3SS_1b is a P. alcalifaciens virulence factor in D. melanogaster. We predict that the T3SS_1b from other Providencia spp., i.e., P. sneebia, P. rettgeri, and P. vermicola (19, 85, 86), will also promote insect infection.

The differential contribution of T3SS_1a and T3SS_1b to vertebrate and invertebrate host colonization, respectively, likely explains their distinct responses to environmental cues. S. Typhimurium T3SS1 and P. alcalifaciens T3SS_1a genes are induced under similar in vitro conditions—late-log phase of growth, aeration, and high salt (60, 87)—and rapidly downregulated following bacterial internalization into eukaryotic cells (60), suggesting that they share commonalities in their regulatory networks. Indeed, P. alcalifaciens T3SS_1a encodes for orthologs of hilA, invF, and sicA (Fig. 2A), three of the main transcriptional regulators of T3SS1 activity in S. Typhimurium (88). In vitro growth conditions that induce T3SS_1b genes remain an enigma as altering the growth phase, temperature, pH, and nutrient availability was unsuccessful. We also did not observe T3SS_1b gene induction upon bacterial infection of mammalian (HCT116) or insect (S2) cells. However, the ∆invA_1b deletion mutant is attenuated in flies, which indicates there are gene induction cues in the insect model that we have not been able to recapitulate in broth or tissue culture cells. T3SS_1b encodes for invF and sicA orthologs, but interestingly, no hilA ortholog is present in this pathogenicity island (Fig. 2A) or elsewhere on the chromosome. Overall, our data support the idea that P. alcalifaciens extended its host range from the natural environment to living organisms by acquiring these two divergent T3SS, each adapted to a target host class: Mammalia (T3SS_1a) or Insecta (T3SS_1b).

Gram-negative bacteria in the genus Providencia are considered opportunistic pathogens. P. alcalifaciens is part of the human gut microbiome (8) and has been isolated from the feces of healthy people and animals (21, 29, 39, 89) and from wastewater treatment plants (90). P. alcalifaciens is also an enteropathogen (31, 32, 39, 40). Based on the available genome sequences and strain source details, we describe a strong link between the subset of P. alcalifaciens strains that harbor T3SS_1a (Fig. 1) and those that cause diarrheal disease in dogs (e.g., strains 2019–04-29292-1-3 and 2019004029290-1-7; Table S1) and humans (e.g., strains 205/92, 2939/90 and RMID 1656011; Table S1). Therefore, it seems that the acquisition of genes encoding a second T3SS on a large plasmid has contributed to the emergence of enteropathogenic P. alcalifaciens strains (45, 91). Also encoded on the 128-kb plasmid of P. alcalifaciens 205/92 are homologs of type III effectors known to be bacterial virulence determinants in mammalian hosts, including SptP, EspG, EspM, SipA_1a, and VopT. SptP from S. enterica has two distinct functions. Its tyrosine phosphatase activity interferes with extracellular-regulated kinase (ERK) MAP kinase pathways in the host cell (92). It also acts as a GTPase activating protein (GAP) for Rac1 and Cdc42 to reverse actin cytoskeletal changes that accompany bacterial entry into epithelial cells (93). The EspG protein family consists of EspG/EspG2 and VirA (94). EspG from enteropathogenic E. coli (EPEC) disrupts the host cell secretory pathway (95). VirA is required for efficient entry of Shigella into epithelial cells and intra- and inter-cellular bacterial spread (96). The EspM/Map/IpgB2 family of effectors are found in EPEC, EHEC, and Shigella and modulate actin dynamics (97). S. enterica SipA and S. flexneri IpaA are functional orthologs that induce a loss of actin stress fibers to promote bacterial entry into non-phagocytic cells (98, 99). In addition to its actin-binding role, SipA is necessary and sufficient for polymorphonuclear leukocyte (PMN) migration across the intestinal epithelium in cellulo and in vivo (100). Of interest, the molecular mass of SipA_1a from a subset of P. alcalifaciens strains (e.g., strains 205/92, 2939–90, 2019–04-28369-1-2, 2019–04-29292-1-3, 2019–04-29034-1-3, 2019–04-2920-1-7, 2019–04-27799-1-2, and 2019–04-3283-1-1; Fig. 1 and 2) is unusually large compared to that of SipA from S. enterica, IpaA from S. flexneri, or CipA from C. violaceum. This is due to an extended C-terminus (Fig. 1B); the biological function of this SipA_1a domain is unknown. VopT from Vibrio parahaemolyticus is a type III translocated toxin that ADP-ribosylates the Ras protein (101). p128kb also encodes for a TcdA/TcdB-type toxin, an exotoxin best characterized in Clostridioides difficile (102). Additionally, the chromosome of P. alcalifaciens 205/92 encodes for homologs of known type III effectors including SteB and SteC from S. Typhimurium and a second SipA (SipA_1b). SteB contributes to S. Enteritidis biofilm formation on plastics (103), but the role of SteB in Salmonella–eukaryotic cell interactions has not yet been deciphered (104). SteC is a kinase that promotes actin cytoskeleton reorganization around the Salmonella-containing vacuole (105, 106). Theoretically, this type III effector repertoire would allow P. alcalifaciens to enter and proliferate within eukaryotic cells. We are currently investigating whether these are bona fide type III effectors and if there are additional type III effectors of P. alcalifaciens.

We report that P. alcalifaciens 205/92 rapidly and efficiently lyses its internalization vacuole in mammalian and insect cells and then proliferates in the eukaryotic cell cytosol. A previous study showed that P. alcalifaciens 101i/59, an invasive isolate, was enclosed within vacuoles and present in the cytosol of Caco-2 cells at 4–6 h p.i. using crystal violet staining (49). Another invasive isolate, 82 A-5778, occupied vacuoles “in close proximity to the nuclear membrane” and the cytosol in Hep-2 cells (46). However, no quantification of vacuolar versus cytosolic residence was provided in these two studies. We speculate that all P. alcalifaciens isolates that harbor T3SS_1a adopt a cytosolic lifestyle inside eukaryotic cells. We further identified mammalian cell responses that are indicative of host sensing of cytosolic bacteria, namely, activation of the caspase-4 (non-canonical) inflammasome, followed by inflammatory cytokine release upon P. alcalifaciens infection of IECs. Caspase-4 also limits the cytosolic proliferation of P. alcalifaciens; whether this is via pyroptotic cell lysis is unknown. There is no evidence for caspase-1-dependent host responses to P. alcalifaciens in colonic epithelial cells, which is in line with studies of other enteric pathogens in human IECs (78, 79).

In addition to inflammasome activation, ubiquitin-mediated autophagy is also an important host innate defense against cytosolic bacteria. Here, we describe that nascent vacuole lysis is partially dependent on T3SS_1a in human IECs, but not human macrophages or insect cells (Fig. 6), indicating host- and cell-type differences in P. alcalifaciens disruption of the surrounding vacuole membrane. Failure of ∆invA_1a bacteria to efficiently lyse the nascent vacuole in IECs results in bacterial killing (Fig. 4). This hints that type III effectors are involved in disrupting the internalization vacuole membrane and preventing the autophagic recognition of P. alcalifaciens in human IECs. Orthologs of type III effectors known to interfere with autophagic recognition (e.g., VirA and IcsB in S. flexneri, SopF in S. Typhimurium, and TssM in Burkholderia spp.) are not present in the genome of P. alcalifaciens 205/92, however. Other known mechanisms used by bacteria to deflect targeting by selective autophagy are LPS and O-antigen modifications (107, 108), enzymatic modification of bacterial outer membrane proteins (108), and bacterial phospholipase manipulation of host cell phospholipids (109–111). Our research efforts are currently directed toward defining how P. alcalifaciens mediates vacuolar escape and resists autophagic detection in mammalian cells.

Prior to our work, rabbits and suckling mice were the two animal models used to study the enteropathogenesis of P. alcalifaciens infection. Murata et al. (31) studied the secretory and inflammatory responses to P. alcalifaciens clinical isolates from a foodborne outbreak in Japan in rabbit ileal loops (31). Histopathological analysis of loops infected with one of these isolates (RIMD1656011) showed extensive mucosal inflammation including neutrophil infiltration within the lamina propria and distortion of the villus architecture. Seven isolates caused a moderate level of fluid accumulation after 20 hours. By contrast, three isolates from diarrheal patients, one isolated in Bangladesh (2939/90) and two in Australia (F90-2004, R90-1475), failed to induce fluid accumulation in the rabbit ileal loop assay after 20 hours (40). Likewise, four P. alcalifaciens isolates from a foodborne outbreak in Kenya did not induce fluid accumulation in rabbit ileal loops over 18 hours in a third study (33). In a suckling mouse model (42), one P. alcalifaciens strain (AH-31) caused diarrhea in 7 of 12 mice, while another (AS-1) did not cause diarrhea in any mice. As an alternative animal model, we considered calves because they are a relevant model for the study of enteric disease associated with human salmonellosis. Published literature indicates Providencia spp. are associated with cattle; antimicrobial-resistant Providencia spp. have been isolated from cow manure (112), feces- and urine-contaminated bedding from yearling calf pens (113), and raw cow’s milk (114), and P. stuartii causes neonatal diarrhea in dairy cows (115). In our studies, fluid accumulation was not observed in calf intestinal loops that were exposed to P. alcalifaciens 205/92 for 12 hours. Histopathological analysis of infected loops also did not show indicators of an acute inflammatory reaction, but mild epithelial damage was observed. Collectively, these studies suggest that either P. alcalifaciens causes non-inflammatory diarrhea, or intestinal loops in rabbits and calves are not an appropriate animal model to study the enteropathogenic properties of this bacterium. It should be noted, however, that the RITARD infection model, which was developed in the early 1980’s to study Vibrio cholerae and enterotoxigenic E. coli enteric infections (116), supports the diarrheagenic nature of P. alcalifaciens clinical isolates (40).

Providencia spp. have been isolated from the hemolymph of wild-caught D. melanogaster (13) and as part of the gut microbiome in Bactrocera dorsalis, the Oriental fruit fly (117), but it is not known if these bacteria are intracellular in insects. Fruit flies are exquisitely sensitive to infection by P. alcalifaciens (our results (13, 81)). In D. melanogaster infections, P. alcalifaciens proliferates rapidly, reaching bacterial loads of ~10⁷–10⁸ CFU per fly by 20–32 h p.i (13, 81). Flies die shortly after (our results (13, 81)). On the host side, Imd-dependent antimicrobial peptides and hemocyte-derived reactive oxygen species are the major branches of insect immunity that are important for fighting infection with P. alcalifaciens (81). In a forward genetics screen using a P. alcalifaciens DSM30120 transposon mutant library, lipopolysaccharide (LPS) and lipoprotein mutants showed reduced virulence in D. melanogaster (81). Even though DSM30120 encodes for T3SS_1b (but not T3SS_1a; Fig. 1C; Table S1), no bacterial mutants in toxins or the T3SS were hit in this screen (81). By contrast, we identified that T3SS_1b, and to a lesser extent T3SS_1a, are virulence factors in D. melanogaster (Fig. 9), which is the first report of P. alcalifaciens T3SSs being virulence determinants in insects. We believe that the different results might be explained by the much lower infection dose used in our study (10–30 CFU vs 1,500 CFU). Considering their high prevalence among Providencia spp. genomes (50), our results suggest that T3SSs are a widespread pathogenicity factor for invertebrate colonization by members of this genus. Furthermore, insects have a well-known role in transmitting clinically relevant pathogens, and Providencia spp. are often resistant to multiple antimicrobials, so studying the insect carriage of bacteria such as P. alcalifaciens may provide information about the role that invertebrate hosts play in the maintenance and transmission of antimicrobial resistance.

While historically viewed as opportunistic pathogens, the Morganella–Proteus–Providencia (MPP) group of organisms (Morganella morganii, Proteus vulgaris, and Providencia species) are increasingly being recognized as emerging causes of multidrug-resistant infections because of inducible chromosomal β-lactamases and a propensity to acquire other resistance determinants. The genomic plasticity of Providencia spp. is noteworthy, and can be seen in the varied lifestyles of different species and strains, ranging from commensal residents of the gastrointestinal tract to assorted pathogens that promote intestinal or extraintestinal illnesses with different clinical consequences. Whole-genome sequencing has provided considerable information about the genomes of non-pathogenic and pathogenic P. alcalifaciens and what genes might contribute to the pathogenicity of this bacterial species (50, 118). From our work, and that of others, it is evident that some P. alcalifaciens strains have gained the ability to enter and proliferate within mammalian cells and cause damage to the gut epithelium and subsequent diarrheal disease. Acquisition of an “invasion-associated” plasmid drove this evolutionary leap. Altogether, our work argues that the importance of P. alcalifaciens as a bona fide enteropathogen should not be ignored and supports its inclusion into systematic surveillance programs.

MATERIALS AND METHODS

Bacterial strains and plasmid construction

P. alcalifaciens 205/92 (Tet^R) served as the WT strain and background for deletion mutants in this study. It was initially isolated from the stool of a 12-year-old boy with watery diarrhea (43). Allelic exchange with a counter-selectable suicide vector harboring sacB (pRE112; Cm^R (119)) was used to generate in-frame deletions of invA_1a and invA_1b. Deletion cassettes were amplified from P. alcalifaciens 205/92 genomic DNA by overlap extension PCR using oligonucleotides listed in Table S2 and ligated into pRE112. The resulting plasmids were electroporated into E. coli SY327λpir for sequence confirmation, then transferred by electroporation into E. coli SM10λpir (Kan^R), followed by conjugation into P. alcalifaciens 205/92 WT (Tet^R). Selection of conjugants was on LB agar containing tetracycline (10 µg/mL) and chloramphenicol (30 µg/mL). The resulting meridiploids were counter-selected by incubating overnight on LB agar containing 1% (wt/vol) tryptone, 0.5% (wt/vol) yeast extract, and 5% (wt/vol) sucrose at 30°C. Sucrose-resistant clones were screened by PCR using primers flanking the recombination region to verify gene deletions. For in trans complementation of ΔinvA_1a, the upstream regulatory region of invF_1a was fused to the open reading frame of invA_1a by overlap extension PCR using the oligonucleotides listed in Table S2 with P. alcalifaciens genomic DNA. The resulting amplicon was then ligated into the BamHI/XmaI sites of pGEN-MCS (120) to generate pGEN-invA_1a, which was then electroporated into P. alcalifaciens ΔinvA_1a bacteria. For fluorescence detection of P. alcalifaciens, bacteria were electroporated with pGEN-DsRed (121), which encodes for the red fluorescent protein variant, DsRed.T3, under the control of the em7 promoter.

To generate transcriptional reporters for genes associated with T3SS_1a, T3SS_1b, and flagella, predicted promoter regions of putative regulators or structural components were amplified with oligonucleotides listed in Table S2 and cloned into pFU35 (122) upstream of the luxCDABE operon. Sequence-confirmed reporter plasmids were electroporated into P. alcalifaciens WT bacteria. Plasmid stability of the pFU series of fusion vectors has been confirmed in E. coli, S. Typhimurium, and Yersinia pseudotuberculosis; in the absence of antibiotic selection, >95% of bacteria retained the plasmids for up to 6 days in liquid culture (122). We expect similar levels of plasmid stability in P. alcalifaciens over the experimental time frames we are using (12 hours in broth and 6 hours in tissue culture cells).

S. Typhimurium SL1344 was the wild-type strain used in this study (123). The S. Typhimurium SL1344 translocator swap mutants, ∆sipC::PasipC_1a and sipC::cipC, have been described previously (51).

Bacterial genome sequencing and assembly

A log-phase culture of P. alcalifaciens 205/92 was centrifuged, and the bacterial pellet was resuspended in 1 x DNA/RNA Shield (Zymo). Bacterial DNA was extracted and sequenced at Plasmidsaurus Inc. (Eugene, Oregon) using Oxford Nanopore Technologies (v14 library prep chemistry, R10.4.1 flow cell, base-called using dna_r10.4.1_e8.2_5 khz_400bps_sup@v4.2.0 model, and primer- and adapter-trimmed) and Illumina (NextSeq 2000, 153-bp paired-end reads). Raw sequencing reads were deposited to ENA (BioProject PRJNA1100810). Raw Illumina reads were adapter- and quality-trimmed using bbduk.sh v38.07, using options “ktrim = r k = 23 mink = 11 hdist = 1 tpe tbo.” Nanopore reads were filtered using filtlong v0.2.1 against the trimmed Illumina reads, with options “--min_length 1000 --keep_percent 90 --target_bases 500000000 –trim --split 500,” selecting approximately 100 x coverage of best Nanopore reads. All scripts used in the analysis can be found in https://github.com/apredeus/P_alcalifaciens.

Resulting filtered Nanopore reads were assembled using Trycycler v0.5.4, which uses manual curation of long-read assemblies to achieve a nearly perfect bacterial genome assembly. To this end, raw Nanopore reads were randomly subsampled to 50 x depth and assembled 10 times each using the following long-read assemblers: flye v2.9.2-b1786, raven v1.8.1, and miniasm 0.3-r179. Following this, 25 assemblies (nine flye, eight raven, and eight miniasm) were selected for further analysis. Assembled contigs were clustered and evaluated; three supported clusters were retained, outlier contigs in each cluster were removed, and reconciled sequences generated. Additionally, Unicycler v0.5.0 was run in the hybrid mode on both Nanopore and Illumina reads to recover small plasmids that long-read approaches could potentially miss because of size selection bias. This allowed the recovery of a small (~4 kb) plasmid missed by long-read-only assembly.

The final assembly consisted of one chromosome and three plasmids. The chromosome was rotated to match the start of RefSeq assembly GCF_002393505.1 (strain FDAARGOS_408, representative P. alcalifaciens genome). Following this, polishing with the trimmed Illumina reads using Polypolish v0.5.0 (124) was done to correct the remaining small indels and single-nucleotide polymorphisms. The resulting assembly was submitted to the NCBI (GenBank Accession No. GCA_038449115.1). To annotate the genome, Bakta v1.9.1 was run with settings “--complete --compliant --genus Providencia --species alcalifaciens --locus-tag PA205 --keep-contig-headers.” To annotate candidate prophage regions, PHASTER online service (https://phaster.ca/, accessed on 15th March, 2024) was used together with the manual curation using individual protein annotations from Bakta.

For the comparative genome analysis, all of the P. alcalifaciens genome assemblies available via Genbank on 10th March 2024 were downloaded as finished assemblies; entries marked as anomalous by the NCBI were excluded. Additionally, representative genomes of P. rustigianii (strain 52579_F01, assembly GCA_900637755.1) and P. rettgeri (strain AR_0082, assembly GCA_003204135.1) were downloaded to be used as potential outgroups. In total, 51 assemblies were downloaded; of these, 13 were marked as “Complete genome,” 28 were marked “Contig,” and another eight were marked “Scaffold” (Table S1). Using snippy v4.6.0 in the contig mode, and our assembly of strain 205/92 as a reference, all other assemblies were mapped and variant-called. Assemblies were evaluated for quality, and assemblies with more than 60% of the reference genome covered were removed (Table S1). Following this, snippy-core utility v4.6.0 was used to create core genome alignment of the remaining genomes.

Pairwise SNP distances were calculated using snp-dist v0.8.2 using the AGCT-only core genome alignment. Constant sites in the full core genome alignment were identified using snp-sites v2.5.1. To determine the plasmid coverage, in silico reads generated by Snippy from assembled contigs (single-end, 250 bp) were mapped to our assembly of P. alcalifaciens strain 205/92 using bwa v0.7.17-r1188, and samtools v1.18 “coverage” command was used to calculate the coverage of individual plasmids. Plasmids with >80% coverage were classified as fully present; plasmids with 40%–80% coverage were classified as partially present.

IQTree v2.2.4 was run using constant sites defined above, and with the options “-redo -ntmax 16 -nt AUTO -st DNA -bb 1000 -alrt 1000.” The best-fit model was selected by the Bayesian information criterion. The resulting tree, plasmid presence, and SNP distances were visualized using R 4.3.3 with packages ggtree (v3.10.1), treeio (v1.26.0), tidytree (v0.4.6), and pheatmap (v1.0.12).

BRIG v0.95 and NCBI blast 2.7.1+ were used to align and produce the circular visualization of the chromosome and the largest plasmid. To be used in BRIG, all of the multi-fasta genome assemblies were converted to a single fasta using a custom script. All scripts used for the analysis and visualization are available at https://github.com/apredeus/P_alcalifaciens.

Bacterial growth curves

P. alcalifaciens was grown overnight in LB–Miller broth (Difco) for 16–18 hours with shaking (220 rpm) at 37°C. Cultures were back-diluted into 10 mL LB–Miller broth in a 125-mL Erlenmeyer flask to a starting optical density at 600 nm (OD₆₀₀) of 0.1 and grown with aeration (220 rpm) at 37°C. OD₆₀₀ was measured hourly in a BioRad SpartSpec Plus spectrophotometer.

Bacterial luminescence

Bacteria were subcultured as described above for 12 hours. Each hour, 150 µL of the subculture was transferred, in duplicate, to a white flat-bottom 96-well polystyrene microplate (Corning Costar) sealed with polyester Axyseal film (Axygen Inc.). Luminescence was measured using a Tecan Infinite M1000 plate reader.

For quantification of bacterial luminescence upon infection of mammalian and insect cells, cells were seeded in 6-well plates: (1) HCT116 cells at 4 × 10⁵ cells/well on rat tail collagen ~40–44 hours prior to infection (2); S2 cells at 2 × 10⁶ cells/well ~ 24 hours prior to infection. Infections with P. alcalifaciens subcultures were as described below. At the required time point, infected monolayers were washed twice with Hanks’ balanced salt solution (HBSS, Corning), collected in 1 mL sterile double-distilled water using a cell scraper (Sarstedt), transferred to a 1.5-mL Eppendorf tube, vortexed, and then centrifuged at 8,000 x g for 2 minutes to pellet bacteria. The supernatant was carefully removed and discarded, and the pellet was resuspended in 100 µL PBS and transferred to a white flat-bottom 96-well polystyrene microplate (Corning Costar). Luminescence was measured using a Tecan Infinite M1000 plate reader.

Motility assays

Overnight cultures of P. alcalifaciens 205/92 WT, S. Typhimurium SL1344 WT, and ΔflgB mutant (60) were inoculated into the center of Petri dishes containing LB plus 0.3% (wt/vol) agar using a sterile pipette tip, piercing approximately half-way through the semi-solid agar. Plates were incubated overnight at 37°C.

Secretion assays and mass spectrometry

Two 10 mL subcultures were grown for 4 hours with shaking for each strain, as described above, and pelleted for 15 minutes at 16,000 x g. Supernatants were pooled and filtered with a 0.22-µm low-protein binding Acrodisc filter (Whatman). Proteins were precipitated in 10% (wt/vol) trichloroacetic overnight at 4°C. Protein precipitates were collected by centrifugation at 16,000 x g for 15 minutes at 4°C, and pellets were washed with cold acetone, dried, and resuspended in 200 µL 1.5X SDS-PAGE sample buffer. Secreted proteins were separated on a 10% SDS-PAGE gel and visualized by Coomassie Brilliant Blue stain (Fisher). For protein identification, secreted proteins were separated on 4%–15% gradient SDS-PAGE gels (BioRad), stained with GelCode Blue (Thermo), and bands of interest were excised and sent to Stanford University Mass Spectrometry (SUMS) facility for protein identification by LC/MS/MS. Following in-gel tryptic digestion, reconstituted samples were analyzed using a nanoAcquity UPLC (Waters) coupled to an Orbitrap Q-Exactive HF-X (RRID:SCR_018703) mass spectrometer.

Tissue culture

HCT116 cells (human colorectal carcinoma epithelial), HeLa (human cervical carcinoma), THP-1 (human monocytes), and J774A.1 (mouse macrophage-like) cells were purchased from ATCC and used within 15 passages of receipt. HCT116 cells were maintained in McCoy’s medium 5A (Iwakata and Grace Modification, Corning), supplemented with 10% (vol/vol) heat-inactivated fetal calf serum (FCS, Invitrogen). HCT116 CASP1^-/- and CASP4^-/- knockout (KO) cells were generated using CRISPR-Cas9 technology by Synthego (www.synthego.com). CASP1^-/- KO clones G15 and I17 and CASP4^-/- KO clones J3 and N8 were derived by single-cell expansion. Premature termination of the respective genes was verified by DNA sequencing. CASP1^-/- KO (clone 1.5) and CASP4^-/- KO (clone 4.15) Caco-2 C2Bbe1 cells (human colorectal adenocarcinoma) and the parental wild-type (WT) cells were provided by Dr Jason Smith (University of Washington) and have been described previously (79). C2BBe1 cells were grown in Dulbecco’s modified Eagle medium (DMEM, 4.5 g/L glucose, Corning) containing 4 mM L-glutamine, 10 µg/mL human transferrin (Sigma), and 10% (vol/vol) heat-inactivated FCS. Where indicated, HCT116 and C2Bbe1 cells were treated with IFNγ (PeproTech) at 50 ng/mL for 16–18 hours. HeLa cells were maintained in Eagle’s minimal essential medium (Corning) containing 2 mM L-glutamine, 1 mM sodium pyruvate, and 10% (vol/vol) heat-inactivated FCS. THP-1 cells were maintained in RPMI 1640 medium (Corning) containing 1 mM sodium pyruvate, 2 mM L-glutamine, 10 mM HEPES, and 10% (vol/vol) heat-inactivated FCS. J774A.1 cells were maintained in DMEM containing 4 mM L-glutamine and 10% (vol/vol) heat-inactivated FCS. All mammalian cell lines were maintained at 37°C in 5% CO₂. Drosophila S2 cells were purchased from Invitrogen and used up to passage number 20. Cells were maintained in Schneiders’ medium (Invitrogen) with penicillin–streptomycin and 10% (vol/vol) heat-inactivated FCS. S2 cells were maintained at 26°C without CO₂. Cells were seeded in 24-well tissue culture-treated plates (Nunc) at the following densities: (i) HCT116 and C2Bbe1, 1 × 10⁵ cells/well ~ 44 hours prior to infection on rat tail collagen (Corning), and growth medium was changed on C2Bbe1 cells to transferrin-free media 4 hours prior to infection, (ii) HeLa, 5 × 10⁴ cells ~24 hours prior to infection, (iii) THP-1, 2.5 × 10⁵ cells/well ~ 48 hours prior to infection in the presence of 200 nM phorbol 12-myristate 13-acetate (PMA, LC Laboratories), and growth medium was replaced with PMA-free media 4 hours prior to infection, (iv) J774A.1, 2 × 10⁵ cells/well ~ 24 hours prior to infection, (iv) S2, 5 × 10⁵ cells/well ~ 24 hours prior to infection, and growth medium was changed to penicillin–streptomycin-free media 4 hours prior to infection.

Gentamicin protection and CHQ resistance assays

To enumerate intracellular bacteria, cells were seeded and infected as described above and subject to a gentamicin protection assay as described (68). P. alcalifaciens strains were subcultured for 3.5–4 hours as detailed above. One mL of subculture was pelleted at 8,000 x g for 90 seconds, resuspended in 1 mL HBSS (Corning), and diluted 1:100 in HCT116 or C2Bbe1 growth media (transferrin-free) or 1:1000 in THP-1 or S2 growth media. Because P. alcalifaciens are poorly motile (Fig. S2), we centrifuged late log-phase cultures onto host cells to promote bacteria–host cell association, a technique typically used for nonmotile or poorly motile bacteria (60, 62, 67). Without this centrifugation step, no P. alcalifaciens were internalized into non-phagocytic (HCT116) cells after 30 minutes of co-incubation (our unpublished results). One milliliter of the diluted culture was centrifuged onto cells at 500 x g for 5 minutes at room temperature (t₀) (multiplicity of infection (MOI) of ~150 for HCT116 and C2Bbe1, ~10 for THP-1, and ~5 for S2), and then monolayers were incubated for a further 25 minutes at 37°C (HCT116, C2Bbe1, and THP-1) or 26°C (S2). Extracellular bacteria were removed at 30 min p.i. by washing three times with 1 mL HBSS and then incubated in growth media containing 100 µg/mL gentamicin at 37°C (HCT116, C2Bbe1, and THP-1) or 26°C (S2). At 1 h p.i., monolayers were washed once with 1X PBS and solubilized in 0.2% (wt/vol) sodium deoxycholate (NaDOC). Internalized bacteria and subculture inocula were serially diluted and plated on LB agar for CFU enumeration. Invasion efficiency was quantified as the number of internalized bacteria/inoculum x 100%. To measure intracellular replication over time, the multiplicity of infection (MOI) for the ∆invA_1a mutant was increased by two- to threefold for HCT116 infections (MOI of 300–600) so that an approximately equivalent number of WT and ∆invA_1a bacteria were internalized at 1 h p.i. Increasing the MOI of ∆invA_1a bacteria up to tenfold did not compensate for its invasion defect in S2 cells, however. Media containing 100 µg/mL gentamicin was added from 30 to 90 min p.i. and then replaced with media containing 10 µg/mL gentamicin for the remaining time. Monolayers were solubilized at 4 , 8 , and 12 h p.i., serially diluted and plated as described above.

To determine the proportion of intracellular bacteria in the cytosol of eukaryotic cells, the CHQ resistance assay was used as previously published (67, 68). The concentration of CHQ was 400 µM for all cell types.

Inside/outside microscopy assay

Extracellular and intracellular bacteria were distinguished by staining with anti-P. alcalifaciens antibodies in the absence of any permeabilizing agents. Cells were seeded on acid-washed 12-mm-diameter glass coverslips (#1.5 thickness, Fisherbrand) in 24-well plates and infected with P. alcalifaciens WT, ∆invA_1a, or ∆invA_1b bacteria carrying pGEN-DsRed.T3. At 1 h and 8 h p.i., monolayers were fixed in 2.5% paraformaldehyde at 37°C for 10 minutes and then washed thrice in PBS. Extracellular bacteria were labeled with rabbit polyclonal anti-P. alcalifaciens antibody (kindly provided by Dr M. John Albert) diluted 1:250 in PBS containing 10% (vol/vol) normal goat serum (PBS-NGS; Invitrogen) for 15 minutes at room temperature. Monolayers were washed thrice in PBS, once in PBS-NGS, and then incubated with goat anti-rabbit Alexa-Fluor 488 secondary antibodies (Invitrogen) diluted 1:300 in PBS-NGS at room temperature for 15 minutes (150 µL per well). After three washes in PBS, host cell nuclei were labeled with Hoechst 33342 for 1 minute (Invitrogen, 1:10,000 dilution in water) and coverslips mounted in Mowiol (Calbiochem) on glass slides. Samples were viewed on a Leica DM4000 upright fluorescence microscope. A bacterium was scored as extracellular if it fluoresced red and green, or intracellular if it fluoresced red only.

Electron microscopy

SEM and TEM sample preparation and imaging was performed at the Franceschi Microscopy and Imaging Center at Washington State University using standard techniques. For SEM, HeLa and HCT116 cells were seeded on Thermanox plastic coverslips (Nunc), infected with P. alcalifaciens as described above, and at 20 min p.i. washed once in HBSS and fixed in 2% paraformaldehyde/2% glutaraldehyde in 0.1 M cacodylate buffer pH 7.2. overnight at 4°C. Glutaraldehyde-fixed cells were dehydrated in an ethanol series and dried at the critical point in CO₂. The samples were sputter coated with platinum/palladium to 2.5 nm thickness and examined on an FEI Quanta 200F scanning electron microscope.

For TEM, HeLa cells were seeded at 2 × 10⁵ cells/well in 6-well plates the day before infection; HCT116 cells were seeded on rat tail collagen at 4 × 10⁵ cells/well in 6-well plates ~ 40 hours prior to infection; THP-1 cells were seeded in the presence of PMA at 1 × 10⁶ cells/well in 6-well plates 2 days prior to infection; S2 cells were seeded at 2 × 10⁶ cells/well in 6-well plates. Monolayers were infected with P. alcalifaciens as described above and at 60–90 min p.i. gently washed thrice with PBS and treated with 0.25% trypsin (Corning) to detach HCT116 and HeLa cells or TrypLE Select (Gibco) to dislodge THP-1 cells. S2 cells were dislodged with a cell scraper (Sarstedt). Cells were collected, centrifuged at 400 x g for 5 minutes, the supernatant discarded, and the cell pellet gently resuspended in fixative. Processing and imaging were done as described previously (66).

Cytotoxicity assays

Prior to infection, the medium was replaced on HCT116 cells and THP-1 cells to phenol-red free RPMI1640 (Corning) containing 10% (vol/vol) heat-inactivated FCS. Infections and subsequent steps were carried out in phenol-red free media. Cells were infected as described above, and supernatants were collected at the indicated times and centrifuged at 500 x g for 5 minutes to pellet cellular debris. Cell-free supernatants were collected and stored at −80°C until analysis. LDH release into the supernatant, a measure of the loss of plasma membrane integrity, was quantified using the Cytotox96 Assay Kit (Promega) according to the manufacturer’s instructions.

Cytokine quantification by ELISA

HCT116 culture supernatants were collected at the indicated times, centrifuged at 500 x g for 5 minutes, and then the cell-free supernatants were stored at −80°C until analysis. IL-18 was quantified by sandwich ELISA as previously described (69). Mouse anti-human IL-18 monoclonal (125–2H) and rat anti-human IL-18 monoclonal (159–12B) biotin were purchased from MBL. Human IL-8 was quantified by DuoSet ELISA (R&D Systems) as per the manufacturer’s instructions.

Immunoblotting

For analysis of whole cell lysates, adherent cells were washed twice in PBS prior to lysis in boiling 1.5 x SDS-PAGE sample buffer. Proteins were separated by SDS-PAGE and transferred to 0.45-µm pore-size nitrocellulose membranes (GE Healthcare Life Sciences). Membranes were blocked at room temperature for 1 hour with Tris-buffered saline (TBS) containing 5% (wt/vol) skim milk powder and 0.1% (vol/vol) Tween-20 (TBST-milk) and then incubated with the following primary antibodies overnight in TBST-milk at 4 ˚C: rabbit polyclonal anti-caspase-1 (A-19) (1:2,000; Santa Cruz Biotechnology), mouse monoclonal anti-caspase-4 (4B9) (1:2,000; MBL), and mouse monoclonal anti-β-actin (8H10D10) (1:20,000; Cell Signaling Technology). Blots were then incubated with anti-rabbit IgG or anti-mouse IgG horseradish peroxidase (HRP)-conjugated secondary antibodies (1:10,000; Cell Signaling) in TBST-milk for 1–2 hours at room temperature, followed by Supersignal West Femto Max Sensitivity ECL Substrate (Thermo). Chemiluminescence was detected using a GE Healthcare AI600 imager.

Quantitative real-time PCR (qPCR)

To evaluate the effect of IFNγ priming on the expression of CASP1, CASP4, CASP5, IL18, and IL1B in HCT116 and C2Bbe1 cells, we used Luminaris Color HiGreen qPCR Master Mix (Thermo Scientific) and a C1000 Touch Thermal Cycler, CFX96 Real-Time System (Bio Rad) with validated oligonucleotide primer pairs as we have described previously (69). Relative gene expression levels were quantified based on quantification cycle (Cq) values and normalized to the reference gene, ribosomal phosphoprotein P0 (RPLP0). The expression of each gene was calculated using the 2^–ΔCq method.

D. melanogaster infections

P. alcalifaciens were grown overnight at 37°C for 16–18 hours with shaking (220 rpm) in LB–Miller broth. Cultures (1 mL) were centrifuged at 8,000 x g for 90 seconds, and the bacterial pellet was washed twice in an equal volume of sterile PBS and diluted 1,000–3,000-fold in sterile PBS. Two-to-7-day old Wolbachia-free adult male w¹¹¹⁸ flies (Bloomington Drosophila Stock Center #5905) were anesthetized with CO₂ and injected with 23 nL of diluted culture (~10–30 CFU/fly) or PBS vehicle control. Flies were maintained on standard cornmeal food at 25°C and 65% relative humidity, and surviving flies were counted every 2–6 h p.i. In each experiment, ~50 flies were injected for each condition, and survival studies were repeated at least twice for each infectious dose (~10 and ~30 CFU/fly).

Bovine ligated jejuno-ileal loop infections

Jersey, Holstein, or cross-bred calves were obtained from North Carolina State University or University of Wisconsin farm herds. Calves were separated from the dam and transferred to AAALAC-approved large animal housing facilities by 1 day of age. Calves were administered either colostrum or colostrum replacer, and adequate passive transfer was estimated by measurement of serum total protein. Calves were treated for 3 days with ceftiofur (4–6 mg/kg SC q24h) and/or flunixin meglumine (50 mg/kg IV q24h) if needed based on the clinical condition upon arrival. Calves were fed milk replacer at 10%–20% body weight per day with ad libitum access to water and hay.

At 3 to 6 weeks of age, calves were anesthetized with intravenous propofol and maintained on isoflurane inhalant for ligated jejuno-ileal loop surgery as previously described (125). Briefly, calves were placed in left lateral recumbency, and a right flank incision was made. Sixteen to thirty-eight 4- to 6-cm loops were tied within the ileum and terminal jejunum, leaving 1-cm spacers between adjacent loops. Prior to inoculation, loop lengths were recorded. Loops injected with vehicle only served as negative controls. Loops were infected individually with 3 mL LB (12-h incubation) or 2 mL PBS (2-h and 8-h incubation) containing approximately 10⁹ CFU of P. alcalifaciens. The intestine was returned to the abdomen, the incision was closed, and the calves were monitored under inhalant anesthesia for the duration of the experiment. At 2 h, 8 h, or 12 h p.i., the incision was opened, and each loop was individually excised. Calves were euthanized by intravenous pentobarbital.

In preparation for ligated loop infections, bacteria were grown shaking (225 rpm) overnight at 37°C in LB–Miller broth. For the 12 h infections, overnight cultures were subcultured 1:100 into LB–Miller broth and further incubated for 3.5–4 hours at 37°C with shaking (225 rpm). Bacteria were washed twice and resuspended in PBS (2 hours or 8 hours) or LB (12 hours) with bacterial normalization based on optical density (OD₆₀₀) for a final inoculation dose of ~10⁹ P. alcalifaciens per loop. The actual inoculum dose was determined by serial dilution and plating. Following loop excision, intestinal fluid and tissue samples were harvested and processed separately. The fluid volume was calculated by excising individual loops and weighing escaped luminal fluid on a sterile Petri dish. Luminal fluid was then transferred to 1 mL sterile PBS to allow for bacterial enumeration. Intestinal tissue was washed twice in sterile PBS to remove non-adherent bacteria and ingesta and then added to 5 mL sterile PBS. For 2 h and 8 h infections, washed tissues were cut in half with one segment treated with gentamicin (50 µg/mL) for 30 minutes at 37 ˚C to quantify intracellular bacteria, and the remaining half was processed to quantify tissue-associated bacteria. After gentamicin treatment, tissues were washed twice with PBS to remove remaining antibiotics. Samples were subsequently homogenized, serially diluted in PBS, and plated for CFU enumeration.

The Institutional Animal Care and Use Committees of North Carolina State University and University of Wisconsin-Madison approved all animal experiments (NCSU protocol numbers 15–047-B and 17–186-B; UW-Madison protocol number V006249). All animal experiments were performed in accordance with the PHS “Guide for the Care and Use of Laboratory Animals” in AAALAC-approved animal facilities.

Histopathology

Intestinal samples were fixed in 10% neutral buffered formalin, processed for paraffin embedding, sectioned (5 µm), and stained with hematoxylin and eosin for histologic analyses. Tissues were assessed and scored by an American College of Veterinary Pathology (ACVP) board-certified pathologist with a scoring system derived from previously published rubrics (126, 127). The following criteria were scored from 0 to 4: lamina propria neutrophil accumulation, submucosal edema, epithelial damage, crypt abscess, villus blunting, and cell death (Table 1).

TABLE 1.

Histopathology scoring system

Score	Lamina propria neutrophil accumulation	Submucosal edema	Epithelial damage	Crypt abscess	Villus blunting	Cell death
0	0–5/hpf	No change (0–5)	No change	None	Normal villi	Absence of cells with morphological features of death
1	6–20/hpf	5–10	Detectable (<10%)	Detectable (<10%)	Few villi mildly blunted	Few sporadic necrotic/apoptotic cells detected
2	21–60/hpf	11–20	Mild (10%–20%)	Mild (10%–20%)	Most villi mildly blunted	Minimal number of necrotic/apoptotic cells detected
3	61–100/hpf	21–40	Moderate (21%–40%)	Moderate (21%–40%)	Most villi moderately blunted	Moderate number of necrotic/apoptotic cells detected
4	>100/hpf	>40	Marked (>40%)	Marked (>40%)	All villi severely blunted	Large number of necrotic/apoptotic cells detected

Statistical analysis

Statistical analysis was performed using GraphPad Prism 10 software. Statistical significance of comparisons between treatment groups was determined using either an unpaired, two-tailed Student’s t test, or for group analysis, using one-way analysis of variance (ANOVA) followed by Dunnett’s multiple-comparison test. For survival curves in flies, a logrank test (Mantel–Cox method) was used to compare two groups.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/iai.00314-24.

Supplemental figures. iai.00314-24-s0001.pdf.

Fig. S1-S4.

Table S1. iai.00314-24-s0002.xlsx.

P. alcalifaciens genome information used for analysis.

Table S2. iai.00314-24-s0003.docx.

Oligos used for cloning.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.