CRISPRi Screening Identifies Essential E. coli Virulence Factors for Placental Barrier Breach in a Maternal–Fetal Infection Model

Abstract

Early-onset neonatal sepsis caused by Escherichia coli (E. coli) threatens neonates’ lives due to the pathogen’s high virulence and multidrug resistance. The mechanisms that enable its placental barrier breach are poorly understood. Using a clinically isolated ST95 ExPEC strain from a neonatal sepsis case, along with a pregnant rat model and an in vitro placental barrier model, we performed CRISPR interference screening. This screen targeted 264 virulence factor genes and identified virulence factors for motility, iron acquisition, hemolysin secretion, and adherence/invasion as critical. We demonstrated that hlyB is essential for uterine infection, and we elucidated a mechanism for ibeA that facilitates syncytial trophoblast cell layer penetration by interacting with the host receptor(s) PSF/VIM to enhance bacterial internalization. Host cells countered ibeA+ E. coli infection via a novel host defense pathway involving upregulation of ASPHD1. This study systematically mapped the virulence factors required for E. coli placental translocation and delineated key host–pathogen interactions.

1. Introduction

Early-onset neonatal sepsis caused by E. coli has become a prominent life-threatening disease because of the pathogen’s high virulence and multidrug resistance. It is estimated that there are up to 5 million cases of neonatal sepsis and about 0.8 million deaths each year [1]. A positive microbial culture from newborn blood was defined as neonatal sepsis. According to postnatal age, it is classified as early- or late-onset sepsis [2]. Pathogens such as Listeria monocytogenes and group B Streptococcus (GBS) can cross the placental barrier and lead to neonatal sepsis [3,4], while incidence of early-onset neonatal sepsis caused by E. coli is increasing. The placenta is a vital barrier against fetal infection, while L. monocytogenes and GBS are known to cross the placenta, the mechanisms for the increasingly prevalent E. coli pathogen are poorly understood. In this study, we focused on the mechanism by which E. coli breaches the placental barrier and infects the fetus.

To screen virulence factors in the complex, multistage infection process of neonatal sepsis caused by E. coli, we introduced a CRISPR interference (CRISPRi) technique in this study. CRISPRi has emerged as a powerful tool for manipulating gene expression in various biological systems [5,6], particularly in the model organism E. coli. This technique leverages the catalytically inactive form of Cas9 (dCas9) to repress gene transcription, facilitating large-scale screening efforts to elucidate gene functions and optimize metabolic pathways [7]. In this study, we aimed to systematically identify virulence factors related to the infection process by applying a CRISPRi pool library against 264 virulence factor genes of a clinically isolated neonatal sepsis pathogen, ST95 ExPEC, in an in vitro placental barrier assay and an SD pregnant rat model to screen virulence factor genes related to the infection process.

Notably, ibeA, which was previously found to be associated with early-onset neonatal sepsis rather than late infections [8], and hlyB, responsible for the exportation of HlyA, which was reported to have a damaging effect on the fetal membrane [9], were found in both screens. The receptor of IbeA, Polypyrimidine Tract-Binding Protein-Associated Splicing Factor (PSF), has been reported to be expressed in placental tissues according to the protein atlas and ibeA has been found to play an important role in penetrating the blood–brain barrier [10], which indicates that the signal pathway in the placenta may mimic the way it acts in the blood–brain barrier; however, the exact role of ibeA in penetrating the placental barrier has not been clarified. We validated the roles of hlyB and ibeA with in vivo and in vitro models and explored how they facilitate placental barrier penetration.

2. Results

2.1. CRISPRi Screen in an In Vitro Placental Barrier Model Identified Genes Facilitating Placental Penetrability

To investigate the virulence factors of ST95 ExPEC that are related to the infection process, we designed and synthesized a library of sgRNAs targeting the CDS start site of 264 virulence genes in ST95 ExPEC (targeting 784 positions, 50 negative control sgRNA). In total, 834 sgRNAs were present in the initial library, representing 100% coverage. The uniformity index was 2.4, indicating that the sgRNAs are uniformly represented in the library. The efficiency of CRISPRi knockdown was examined using a GFP reporter, and the fluorescence intensity was significantly reduced 3 h after 0.6 mM IPTG addition (Figure 1A,B); this induction condition was used in the screens.

Figure 1.

CRISPRi screening reveals virulence factor genes in in vitro placental barrier assay. Visualization of the plasmid structure of the GFP reporter (GFP vector) used to test CRISPRi efficiency and dCas9 for CRISPR interference (pdCas9-E4 vector) (A). CRISPRi inhibits GFP expression significantly in the IPTG-treated group compared to the untreated group, guided by sgRNA targeting the transcription start site (B), *** stands for p < 0.001, **** stands for p < 0.0001. Principal Component Analysis (PCA) of sgRNA distributions from different samples, with most samples clustered together (C). Screening using the in vitro placental barrier assay (D) showed that sgRNAs targeting flhB, entB, hlyB, ibeA, cheA, csgE, csgG, fepD, fepE, fimI, flhA, flhE, fliG, fliO, gspF, hofB, iucB, motA, neuC, papJ, ppdD, sfaF, ycfz were significantly downregulated in the lower chamber of the transwell culture compared to the upper chamber of the transwell culture, indicating that these virulence factor genes are involved in facilitating syncytial trophoblast cell layer penetration. Volcano plot of sgRNA distribution in the lower chamber versus the upper chamber for the IPTG-treated group in the in vitro placental barrier assay (E). Gene Ontology of in vitro placental barrier penetration genes (F).

We then screened virulence factor genes in an SD pregnant rat model and in an in vitro placental barrier model. For most samples, sgRNA distribution clustered together (Figure 1C), as analyzed by Principal Component Analysis (PCA) and we excluded significant sgRNA types in the uninduced samples compared to control samples in the differential gene analysis.

In the in vitro screening experiments, we compared lower-chamber sgRNAs to upper-chamber sgRNAs in the transwell culture and used robust ranking aggregation to perform differential gene analysis (Figure 1D,E). A total of 23 virulence factor genes including flhB, entB, hlyB, ibeA, cheA, csgE, csgG, fepD, fepE, fimI, flhA, flhE, fliG, fliO, gspF, hofB, iucB, motA, neuC, papJ, ppdD, sfaF, and ycfz were found to be associated with crossing the syncytial trophoblast cell layer, the Gene Ontology of which is described in Figure 1F.

2.2. CRISPRi Screen in SD Pregnant Model Identified Genes Related to the Infection Process

In the in vivo screen experiments, we compared tissue sgRNAs with vaginal sgRNAs, using robust ranking aggregation to perform differential gene analysis, and found that sgRNAs targeting flhB, entB, and hlyB were significantly downregulated in the amniotic fluid, placenta, and umbilical cord compared to the vagina in the IPTG groups (Figure 2A–C,G), while the p value of sgRNAs targeting neuC in these three tissues were approaching 0.05 compared to the vagina in the IPTG groups. The sgRNAs targeting flhB, entB, hlyB, and neuC were also significantly downregulated in fetal blood compared to the vagina in the IPTG groups. In addition, sgRNAs targeting ibeA and 80 other virulence factor genes were found to be significantly downregulated in fetal blood compared to vaginal samples (Figure 2D), the Gene Ontology of which is described in Figure 2E. The significantly downregulated sgRNA targeting genes in the amniotic fluid, placenta, and umbilical cord indicate that initial infection reliance on flhB, entB, and hlyB may also require neuC, but more virulence factor genes are needed to infect the fetus. The sgRNAs targeting flhB, entB, hlyB, neuC, ibeA, csgE, csgG, flhA, flhE, fliG, and iucB found on the in vitro screen (lower-chamber culture medium vs. upper chamber) were also significantly downregulated in the in vivo screen (fetal blood vs. vagina) (Figure 2F) [11], which highlighted their significant role in placental penetrability. It is noteworthy that maternal blood samples from the IPTG groups failed to produce any bacterial clones, while uninduced groups grew clones. Considering that samples of fetal blood both grew clones, bacteremia in the mother may be dispensable for ascending infection via the vaginal route.

Figure 2.

CRISPRi screening reveals virulence factor genes in an SD pregnant rat model. Volcano plot of sgRNA distribution for the IPTG-treated group in the amniotic fluid versus vagina in the IPTG-treated group (A), placenta versus vagina in the IPTG-treated group (B), umbilical cord versus vagina in the IPTG-treated group (C), and fetal blood versus vagina in the IPTG-treated group (D), the dashed lines separate significant downregulated (left), upregulated (right) sgRNAs. Gene Ontology of fetal blood versus vaginal genes (E). Venn diagrams (F) show that all samples shared common virulence factor genes: flhB, entB, hlyB. The overlapping genes in the fetal blood and in vitro screen were flhB, entB, hlyB, neuC, ibeA, csgE, csgG, flhA, flhE, fliG, iucB. In vivo screening (G) shows flhB, entB, hlyB are important for E. coli infection in the amniotic fluid, placenta and umbilical cord, while flhB, entB, hlyB, neuC, ibeA and 80 other virulence factor genes are important for E. coli infection in the fetal blood, indicating that both overlapping and distinct sets of virulence factor genes are involved in different stages of infection, arrows show the ascending infection process of ST95 ExPEC.

2.3. ST95 ExPEC Infection Leads to Neonatal Sepsis

We examined the infectivity of ST95 ExPEC in a pregnant SD rat model. After ST95 ExPEC inoculation in the vagina at E16, high levels of ST95 ExPEC (examined by MLST) were detected in the fetal blood, amniotic fluid, umbilical cord, and placenta at E20, indicating that vaginal inoculation with ST95 ExPEC led to fetal infection, further demonstrating that this ST95 ExPEC strain is a neonatal sepsis pathogen (Figure 3A).

Figure 3.

E. coli of different genetic backgrounds exhibits distinct infectivity. E. coli (10⁸ CFU E. coli was inoculated vaginally into pregnant SD rats at E16. Dams were euthanized at E20, and the bacterial loads (CFUs) in the fetal blood, placenta, amniotic fluid, and umbilical cord were quantified (A). ST95 ExPEC infects the fetal blood, placenta, amniotic fluid, and umbilical cord of pregnant rats, △hlyB only colonizes the vagina and cannot lead to intra-uterine infection; ibeA expressing TOP10 E. coli only colonizes the vagina and infects the placenta, while GFP expressing TOP10 E. coli only colonizes the vagina. The values represent the number of animals with a tissue CFU count between 20 and 100 divided by the total number of animals in the group. * stands for CFU count > 1000 or above the maximum limit. HE staining of the maternal side (B) and embryo side (C) of the placenta vaginally treated with 10⁸ CFUs of ST95 ExPEC at E16 in pregnant SD rats shows infiltrated neutrophils (arrows) at E20. HE staining of the maternal side of the placenta vaginally treated with 10⁸ CFU △hlyB (D), ibeA (F), GFP (H), control (J), and the embryo side of the placenta vaginally treated with 10⁸ CFU △hlyB (E), ibeA (G), GFP (I), and control (K) in pregnant SD rats showed no infiltration of neutrophils at E20. Scale bar: 50 μm.

HE staining revealed that ST95 ExPEC infection induced neutrophil infiltration in the maternal and embryo side of the placenta (Figure 3B,C), while the placenta of the △hlyB group (Figure 3D,E), ibeA group (Figure 3F,G), GFP group (Figure 3H,I), and control group (Figure 3J,K) showed no signs of inflammation. No signs of inflammation were found in the umbilical cord in any of the groups.

The syncytial (fused) BeWo cell layer forms a more effective barrier against ST95 ExPEC penetration than the unfused BeWo cell layer or fibronectin-coated transwell membrane (Figure 4A), as demonstrated by the in vitro placental barrier assay. This indicates that syncytial trophoblasts are a more effective barrier against ST95 ExPEC in vitro; however, they could not prevent fetal infection in vivo.

Figure 4.

In vitro placental barrier assays show virulence factor genes hlyA, hlyB and ibeA facilitate syncytial trophoblast penetration and LDH assay shows that the cytotoxic effect of hlyB relies on hlyA. An in vitro placental barrier assay was performed to compare the penetration rate (lower-chamber CFUs/upper-chamber CFUs) of ST95 ExPEC to fused BeWo cells induced by forskolin, unfused BeWo monolayer, or fibronectin-coated transwell membrane, (A) which indicated syncytial trophoblast layers formed a stronger barrier against ST95 ExPEC penetration. The penetration rate of ST95 ExPEC and △hlyA, △hlyB, △hlyA△hlyB (B) indicated hlyB-facilitated syncytial trophoblast relies on hlyA. The WST-8 LDH Activity Assay (C) was applied to measure LDH activity in the supernatant and cell lysis of ST95 ExPEC-, △hlyA-, △hlyB-, and △hlyA△hlyB-treated fused BeWo cells (5 h, multiplicity of infection (MOI) of 100). LDH release was calculated as the total activity of the supernatant LDH/total activity of the cell lysis LDH, indicated the cell toxicity of HlyB on syncytial trophoblast cells relied on HlyA. The penetration rate of ST95 ExPEC and △ibeA (D), TOP10 E. coli that overexpressed ibeA and GFP (E) to fused BeWo cells indicated ibeA facilitated syncytial trophoblast penetration. The penetration rate of ST95 ExPEC to fused BeWo cells transfected with PSF RNAi (F), VIM RNAi (G) or NC RNAi indicated IbeA receptors PSF and VIM affect ST95 ExPEC penetration. The penetration rate of △ibeA (H) or Top10 E. coli that overexpressed GFP (I) to fused BeWo cells transfected with PSF RNAi, VIM RNAi, or NC RNAi indicated the penetration effect of IbeA relies on IbeA receptor PSD and VIM. Fused BeWo cells were induced by forskolin, 4 h after bacterial inoculation (10⁶ CFU for TOP10, 10⁵ CFU for ST95 ExPEC) at the upper chamber of the transwell culture. ns stands for not significant, p > 0.05, * stands for p < 0.05, ** stands for p < 0.01, **** stands for p < 0.0001.

2.4. hlyB Facilitates Infection by Interacting with hlyA

We examined △hlyB infectivity in a pregnant SD rat model. After ST95 ExPEC inoculation in the vagina at E16, no uterine infection was detected at E20, whereas a high level of △hlyB clones was detected in the vagina (Figure 3A).

hlyB is reported to be responsible for HlyA secretion in E. coli [11]. We then examined the genetic interactions in in vitro models. In vitro placental barrier assays showed a significant reduction in the syncytial trophoblast penetrability of △hlyA, △hlyB and △hlyA△hlyB compared to ST95 ExPEC, while △hlyA△hlyB showed no further significant reduction compared to △hlyA or △hlyB (Figure 4B).

Measuring LDH (Lactate Dehydrogenase) release from the supernatant is an approach to evaluate cell damage and death. LDH release assays showed a significant increase in LDH release in ST95 ExPEC-stimulated syncytial trophoblasts compared to control cells (Figure 4C), but a significant reduction in LDH release in △hlyA-, △hlyB- and △hlyA△hlyB-stimulated syncytial trophoblasts compared to ST95 ExPEC-stimulated cells, whereas △hlyA△hlyB showed no further significant reduction compared to △hlyA or △hlyB. These results indicate hlyA and hlyB interact genetically in the same pathway of ST95 ExPEC penetrability and cell damage effects on syncytial trophoblast cells, as reported in a previous paper [11].

2.5. The Virulence Factor ibeA Facilitates Placental Infection Through Interactions with Host Receptors PSF and VIM

Vaginal inoculation of ibeA expressing TOP10 E. coli led to placental infection, whereas vaginal inoculation of GFP expressing TOP10 control E. coli only colonized the vagina (Figure 3A). In vitro placental barrier assays showed that in the ST95 ExPEC genetic background, the penetration rate of the ibeA knockout strain was significantly lower than that of the wild-type ST95 ExPEC (Figure 4D), while in the TOP10 E. coli genetic background, the penetration rate of ibeA-expressing E. coli was significantly higher than that of the GFP-expressing control E. coli (Figure 4E), which indicates ibeA enhances E. coli penetration into the syncytial trophoblast BeWo cell layer.

We further explored the role of the known IbeA receptors PSF and VIM in an in vitro placental barrier model. RNAi knockdown of the IbeA receptor PSF significantly lowered the ST95 ExPEC penetration rate (Figure 4F), whereas RNAi knockdown of vimentin protein receptor VIM significantly enhanced ST95 ExPEC penetration (Figure 4G), which indicates IbeA receptors PSF and VIM act differently in BeWo cells. RNAi knockdown of the IbeA receptor PSF or VIM did not affect △ibeA strain and TOP10 GFP penetration, indicating that the RNAi effect on IbeA receptors relies on the expression of ibeA in E. coli (Figure 4H,I).

ST95 ExPEC stimulation of fused BeWo cells significantly inhibited VIM expression, while cytosolic PSF levels did not change significantly, which is consistent with the result that VIM RNAi increased the penetration rate of ST95 ExPEC into syncytial trophoblast BeWo cells. In addition to being located in the nucleus, PSF also distributed in the plasma membrane (Supplementary Figure S1K).

2.6. E. coli with ibeA Penetrate Syncytial Trophoblast Cell Layer by Enhanced Internalization

In addition to cell contacts, entering cells is another pathway that E. coli uses to penetrate the cell layers. Syncytialization significantly reduced ST95 ExPEC and GFP TOP10 E. coli internalization, indicating that fewer bacteria had entered the cells (Figure 5A,B).

Figure 5.

Gentamicin protection assay shows bacterial internalization is affected by syncytialization, virulence factor gene ibeA, and its receptors PSF and VIM. A gentamicin protection assay was performed to compare the number of ST95 ExPEC (A) and GFP-expressing TOP10 E. coli bacteria (B) internalized by fused BeWo cells and unfused BeWo cells, indicating syncytialization reduces bacterial internalization. The number of ST95 ExPEC and △ibeA internalized by unfused BeWo cells (C) or fused BeWo cells (D) and the number of TOP10 E. coli, constitutively expressing GFP or ibeA, internalized by unfused BeWo cells (E) or fused BeWo cells (F) indicated ibeA facilitated bacterial internalization. The number of ST95 ExPEC internalized by fused BeWo cells (G) or unfused BeWo cells (H) treated with 100 μM wortmannin for 1 h, ST95 ExPEC internalized by unfused BeWo cells treated with 33 μM nocodazole (I) for 1 h, and ST95 ExPEC internalized by fused BeWo cells (J) and unfused BeWo cells (K) treated with 25 μM Vimentin-IN-1 for 1 h and BeWo cells treated with the same volume of DMSO, indicated PI3K, microtubule and vimentin affects ST95 ExPEC internalization. The number of ST95 ExPEC internalized by fused BeWo cells transfected with VIM RNAi (L), PSF RNAi-3 (N) or NC RNAi, ST95 ExPEC internalized by unfused BeWo cells (M) transfected with PSF RNAi-3 or NC RNAi, and △ibeA (O) internalized by fused BeWo cells transfected with PSF RNAi-3, VIM RNAi or NC RNAi indicated that the enhanced bacterial internalization effect of IbeA receptors PSF and VIM relies on IbeA. ns stands for not significant, p > 0.05, * stands for p < 0.05, ** stands for p < 0.01, *** stands for p < 0.001, **** stands for p < 0.0001.

The expression of ibeA enhanced the ability of BeWo cells to internalize bacteria, as demonstrated by strains in both the ST95 ExPEC background (Figure 5C,D) and TOP10 E. coli background (Figure 5E,F), whereas ibeA did not alter tight junction protein expression (Supplementary Figure S1B). This indicates that ibeA enhances syncytial trophoblast penetration by enhancing bacterial internalization.

Inhibitors targeting microtubulin (nocodazole) and macropinocytosis (wortmannin) significantly reduce the internalization of bacteria [12]. To study the role of internalization in ST95 ExPEC infection, these inhibitors were used in gentamicin protection assays.

Treatment with the PI3K inhibitor wortmannin significantly reduced ST95 ExPEC internalization in both fused and unfused BeWo cells (Figure 5G,H), and wortmannin treatment also enhanced ST95 ExPEC penetration, which may be due to cell death induced by wortmannin (Supplementary Figure S1F). The microtubule inhibitor nocodazole inhibited the internalization of ExPEC in unfused BeWo cells (Figure 5I) but did not affect bacterial internalization in fused cells.

Vimentin-IN-1, a vimentin inhibitor, also increased ST95 ExPEC internalization in both fused and unfused BeWo cells (Figure 5J,K), whereas VIM RNAi also increased ST95 ExPEC internalization in fused BeWo cells (Figure 5L), which coordinated with the penetration enhancement in VIM RNAi (Figure 4G). RNAi knockdown of the other IbeA receptor, PSF, increased ST95 ExPEC internalization in both fused and unfused BeWo cells (Figure 5M,N); however, its penetration rate decreased significantly (Figure 4F). RNAi knockdown of the IbeA receptor PSF or VIM did not affect △ibeA strain internalization, indicating that the RNAi effect on IbeA receptors relied on the expression of ibeA in ST95 ExPEC (Figure 5O).

2.7. Host Transcriptional Response to ibeA+ E. coli

To investigate the transcriptional change following IbeA stimulation, we performed RNA-seq of a total of six groups of BeWo cells, with each group containing three individual replicates. The unfused BeWo cells, fused BeWo cells, fused BeWo cells treated with ST95 ExPEC, fused BeWo cells treated with △ibeA, fused BeWo cells treated with TOP10 E. coli expressing GFP, fused BeWo cells treated with TOP10 E. coli expressing ibeA and the co-cultivation time of E. coli and BeWo cells was 6 h. We intersected the differential genes of cells stimulated by strains carrying ibeA and the control gene in both ST95 ExPEC and TOP10 E. coli (Figure 6A,B). The only gene that was upregulated upon ibeA stimulation in both E. coli strains was ASPHD1 (Figure 6C), whereas BNIP3P11, LINC02014, NPFF, ALOX15P1, TNFRSF13C, and CTAGE6 were downregulated. ASPHD1 is a transmembrane protein that is located in the plasma membrane. It is a member of the aspartate beta-hydroxylase family and is associated with colon cancer. However, its function remains to be explored [13]. RNAi knockdown of ASPHD1 significantly enhanced ST95 ExPEC internalization (Figure 6D,E) in both fused and unfused BeWo cells. Doxorubicin, an inhibitor of ASPHD1 [14], significantly enhanced ST95 ExPEC penetration into fused BeWo cells (Figure 6F). Upregulation of ASPHD1 upon IbeA stimulation may be a defensive adaptation to ST95 ExPEC invasion.

Figure 6.

RNA-seq reveals ibeA upregulates ASPHD1 as a defensive strategy against ibeA+ E. coli. Volcano plot of mRNA distribution in ibeA versus GFP (A) (389 significant genes, 167 downregulated, 222 upregulated) and ST95 ExPEC versus (398 significant genes, 251 downregulated, 147 upregulated) △ibeA-treated fused BeWo cells (B). Venn diagram showing the intersection of upregulated genes of ibeA versus GFP and ST95 ExPEC versus △ibeA is ASPHD1 (C). A gentamicin protection assay was performed to compare the number of ST95 ExPEC internalized by unfused BeWo cells (D) or fused BeWo cells (E) transfected with ASPHD1 RNAi and BeWo cells transfected with NC RNAi indicating ASPHD1 knockdown enhanced bacterial internalization. In vitro placental barrier assay was performed to compare the penetration rate (lower-chamber CFUs/upper-chamber CFUs) of ST95 ExPEC to fused BeWo cells treated with 5 μM doxorubicin (F) for an hour and fused BeWo cells treated with the same volume of DMSO 4 h after bacterial inoculation (10⁵ CFUs) in the upper chamber of the transwell culture, indicating ASPHD1 activity was required to reduce ST95 ExPEC penetration. * stands for p < 0.05, ** stands for p < 0.01. Mechanism of how ibeA facilitates ST95 ExPEC penetration of the placental barrier by enhancing bacterial internalization by interacting host receptors PSF and VIM, and in response to ibeA+ E. coli infection, host upregulates ASPHD1 to lower bacterial internalization to reduce infection (G), ? stands for may be through CREB, ibeA leads to transcriptional upregulation of ASPHD1.

3. Discussion

3.1. Virulence Factors Related to Infection

The ST95 ExPEC strain isolated from a neonatal sepsis patient is a key pathogen in neonatal sepsis. In our study, we systematically mapped the functions of virulence factors employed by ST95 ExPEC to breach the placental barrier, highlighting the stage-specific requirements for fetal infection. The infection may originate from vaginally colonizing bacteria, breach the placental barrier, and infect the fetus [15,16]. The in vivo experiments showed consistent reliance on the virulence factor genes flhB, entB, and hlyB in amniotic fluid, placental and umbilical cord infection. flhB is responsible for bacterial motility, entB is responsible for iron acquisition, and hlyB is responsible for the exportation of hemolysis factor HlyA. The loss of hlyB is sufficient to prevent infection. Further infection of fetal blood requires more virulence factor genes.

The intersection of in vivo screening (fetal blood) and in vitro screening (placental barrier assay) revealed that the virulence factor genes responsible for bacterial motility (flhB, flhA, flhE, fliG) [17,18,19,20,21], iron acquisition (entB, iucB) [22,23], biofilm formation and adherence (neuC, csgE, csgG) [24,25,26], blood–brain barrier penetration (ibeA) [10], and hemolysis (hlyB) [11] are essential for penetration of the placental barrier and fetal infection. E. coli carrying these virulence factors should be considered as a potential neonatal infector.

In another in vitro screen of the virulence factor genes that enhance bacterial internalization, we isolated ibeA, the results of which will be reported separately.

Our study introduced an innovative approach by combining a CRISPRi library with an in vivo model to efficiently identify stage-specific virulence factors. This was achieved by designing sgRNAs that selectively target virulence genes, avoiding those essential for growth, and then analyzing the bacteria isolated from different host tissues to identify genes critical for each step of the infection process. From a methodological standpoint, our research established a robust framework for studying barrier penetration in vivo. The CRISPRi screening platform allows for precise functional annotation of virulence genes directly in the relevant host environment, moving beyond correlative genomic studies to causal validation.

3.2. ST95 ExPEC Reduces Tight Junction Protein Expression

While the primary mechanism identified for ibeA is enhanced internalization, barrier integrity was also found to be important for ST95 ExPEC penetration. The barrier effect of syncytial trophoblasts may be derived from tight junctions and fused cell states that reduce the number and enhance the strength of cell contacts that allow bacteria to pass through. The expression level of the tight junction protein occludin decreased significantly after syncytialization, which may result from a reduction in the number of tight junctions after syncytialization, while the expression level of ZO-1 did not change significantly. Occludin levels decreased significantly after 6 h of ST95 ExPEC stimulation (10⁷ CFUs) compared with sterile fused BeWo cells (Supplementary Figure S1A). However, E. coli carrying ibeA did not affect ZO-1 or occludin expression levels compared to control E. coli (Supplementary Figure S1B). Although PSF and VIM RNAi significantly lowered their target gene expression (Supplementary Figure S1C,D), they did not affect ZO-1 and occludin expression levels. The penetration effect of ibeA may not result from tight junction disruptions.

3.3. The Mechanism of ibeA

As one of the 264 virulence factors, ibeA has been reported to be an essential virulence factor that facilitates E. coli K1 to cross the blood–brain barrier [10]. IbeA has been characterized as a homo-dimeric enzyme within the class of FAD-dependent oxidoreductases, possessing the ability to associate with two flavin adenine dinucleotide (FAD) molecules per homodimer [27]. The structural configuration of IbeA, specifically its homo-dimeric nature and affinity for dual FAD molecules, may underpin its functional role in conferring resistance to hydrogen peroxide (H₂O₂) [28]. The IbeA protein significantly enhances the invasive capability of E. coli by interacting with specific host cell receptors, including PSF, p54nrb, vimentin, and Caspr1, which interact with the IbeA protein and facilitate E. coli invasion [10,29,30]. Notably, PSF and p54nrb are RNA-binding proteins with considerable sequence similarity, and their overexpression has been shown to augment E. coli K1 invasion, whereas addition of their recombinant protein forms can significantly inhibit this process. The other receptor, vimentin gene VIM, acts in concert with PSF [29]. Although ibeA was reported to act with PSF and VIM through the NF-κB pathway in HBMECs, we used CAPE, the same NF-κB inhibitor in a previous paper, and the in vitro placental barrier assays showed that CAPE does not affect ST95 ExPEC penetration into syncytial trophoblast cells (Supplementary Figure S1E), indicating that an NF-κB-independent mechanism and bacterial internalization may explain the cases in BeWo cells, as demonstrated in the paper.

A novel finding of our study was the discovery of ASPHD1 as a novel host defense factor. The specific upregulation of ASPHD1 in response to ibeA+ E. coli and its functional role in limiting bacterial internalization suggest a previously unrecognized innate immune surveillance system at the placental interface. The IbeA-ASPHD1 axis represents a refined molecular arms race: the pathogen utilizes ibeA to exploit host receptors for invasion, whereas the host counters by deploying ASPHD1 to safeguard against this specific threat. The identification of ASPHD1 opens new avenues for exploring host-directed therapies to bolster placental defense.

4. Materials and Methods

4.1. Ethics Approval

The study was conducted in accordance with the Declaration of Helsinki. The acquisition of clinical ST95 ExPEC strains (Approval No. 2019-k-350, 1 October 2019) was reviewed and approved by the Institutional Review Board (IRB) of Capital Medical University Affiliated Children’s Hospital, Beijing, China. The animal study (Approval No. DWLL2025017, 10 July 2025) was reviewed and approved by the Ethics Committee of the Children’s Hospital, Capital Institute of Pediatrics, Beijing, China. All methods were performed by relevant guidelines and regulation.

4.2. Bacterial Strain

Nine ST95 ExPEC strains were isolated from the mother and her babies of dizygotic twins who suffered early-onset neonatal sepsis and were delivered through a cesarean section. Whole-genome sequencing of the strains was performed and they were found to have a 99.99% sequence similarity. The strain used for the following experiments contained 264 virulence factors, which were analyzed according to the ecoli_vf database and are listed in the Supplementary File.

To generate gene knockout strains, we used the λ Red recombinase system to substitute the coding regions of ibeA, hlyA and hlyB with the KanR gene in the ST95 ExPEC genetic background. Since the hlyA and hlyB genes are adjacent in the genome, we substituted the whole continuous region of hlyA and hlyB with KanR to construct the △hlyA△hlyB double mutant strain.

To generate gene-overexpressing strains, we first constructed the ibeA::flag expressing plasmid driven by the J23119 promoter, which enabled the constitutive expression of virulence factor genes. The plasmid structure is shown in Supplementary Figure S1I,J. We transduced ibeA-expressing plasmid into TOP10 chemically competent E. coli cells in the experimental group and GFP in the control group.

4.3. sgRNA Design, Vector Construction, dCas9 Expression Induction and CRISPRi Efficiency Test

To use CRISPRi to repress gene expression, the pdCas9-E4 vector was used in this study. The plasmid structure is shown in Figure 1A. sgRNA was designed to target the transcription start site of each virulence factor gene. A total of 785 sgRNAs targeting virulence factor genes and 50 control sgRNAs were designed. All sgRNA sequences are listed in the Supplementary File. After we synthesized sgRNA primers, ligated to the pdCas9-E4 vector, and next-generation sequencing was performed, the library plasmids were electroporated into ST95 ExPEC cells, then bacteria were inoculated in LB with kanamycin and cultured at 37 °C and 200 rpm to an OD600 of ~0.6; IPTG was added to the culture medium to 0.6 mM and cultured at 37 °C and 200 rpm for 3 h to induce dCas9 expression; then, bacteria were pelleted and applied in the in vitro and in vivo experiments. To validate the CRISPRi efficiency, we transduced a GFP reporter vector (containing a chloramphenicol-resistant gene) and a pdCas9-E4 vector (containing a kanamycin-resistant gene) that contained an sgRNA targeting the transcription start site of GFP into ST95 ExPEC, then cultured it in LB with kanamycin and chloramphenicol at 37 °C to reach OD600~0.6 and IPTG was added to 0.6 mM to induce dCas9 expression for 3 h; then, GFP fluorescence intensity was measured. The negative control was ST95 ExPEC transduced with a pdCas9-E4 vector (cultured in LB with kanamycin), and the positive control was transduced with a GFP vector and a pdCas9-E4 vector that did not contain sgRNA. All groups were cultured under the same conditions and simultaneously induced at the same time.

4.4. Cell Culture

To study the placental barrier in vitro, we selected BeWo, a cell line exhibiting epithelial morphology isolated from the placenta of a patient with choriocarcinoma. It is an immortal cell line that is capable of forming syncytial barriers induced by forskolin. It is a cell line with a slow growth rate; the F-12K culture medium should be changed every other day and passaged every six days. To induce cells into a syncytial state (fused), BeWo cells were seeded on day 0, and the culture medium was changed to F-12K containing 10% FBS and 50 ng/mL EGF. On day 3, the culture medium was replaced with F-12K containing 10% FBS, 50 ng/mL EGF, and 50 μmol/L forskolin. On day 5, the culture medium was changed to F-12K with 10% FBS. To initiate the transwell and bacterial internalization assays, the culture medium was replaced with F-12K without antibiotics on day 7.

4.5. Bacterial Penetration Transwell Assay

To study the penetrability of E. coli into the syncytial trophoblast cell layer, we established a transwell assay based on the BeWo cell line. A 6.5 mm 24 well 3.0 μm transwell was selected for this transwell assay. Before seeding the cells, 50 μL 10 μg/mL fibronectin was coated to the transwells for 2 h at 37 °C and then the transwells were washed with DPBS to increase the adherence of cells. A total of 5 × 10⁵ cells/cm² was seeded on day 0, and syncytialization was induced according to the protocol mentioned above. For the unfused group, the culture medium was changed every alternate day. For strains in the ST95 ExPEC background, 10⁵ CFUs were added to the upper chamber, whereas 10⁶ CFUs were added to the TOP10 E. coli strains. The sample CFUs in both chambers were checked at 4 h by plating on LB plates and culturing at 37 °C overnight. There were three samples in each group. The penetration rate was calculated using the lower-chamber CFUs/upper-chamber CFUs and each group had three replicates.

4.6. LDH Release Assay

To study the cell damage effect of hemolysin-expressing E. coli on syncytial trophoblast cells, we used the LDH Activity Assay Kit with WST-8 (C0018S) to measure LDH activity in the supernatant and cell lysis of syncytial (fused) BeWo cell cultures 5 h after E. coli stimulation (multiplicity of infection (MOI) of 100). There were three samples in each group. LDH release% was calculated as total supernatant LDH activity/total cell lysis LDH activity × 100%.

4.7. Bacterial Internalization Assay

A gentamicin protection assay was performed to measure the ability of BeWo cells to internalize E. coli; 96-well plates were selected for the bacterial internalization assay, and 10,000 cells were seeded in each well. Syncytialization was induced as previously described. For the RNAi groups, after the first 5 days of syncytialization, siRNA was transfected into cells using RNAimax and the culture medium was changed the next day. On the third day after RNAi transfection, the culture medium was replaced with 10% FBS F-12K without antibiotics. A total of 10⁷ CFUs of E. coli cells were added to the wells, and after cultivation in a CO₂ incubator (5% CO₂ content) at 37 °C for 1 h, the cells were washed with an F-12K medium containing gentamicin. The cells were cultured in a 10% FBS F-12K gentamicin medium for 1 h. The cells were washed with F-12K without gentamicin, lysed with sterile ddH₂O, and plated onto LB plates. The CFUs were calculated and there were three samples in each group. All of the washing processes were performed gently to prevent the cells from falling off the plates.

4.8. In Vitro Screen Assay

To screen for the virulence factor genes that facilitate syncytial trophoblast cell layer penetration, we applied CRISPRi screening in an in vitro placental barrier model. After 7 days of syncytialization induction, the IPTG-induced CRISPRi library ST95 ExPEC (10⁵ CFUs) was added to the upper layer of the transwell. Three samples were analyzed in the IPTG-induced group, and two in the uninduced control group. After 4 h of bacterial seeding, the culture medium of the upper and lower layers of the transwell was transferred to liquid LB containing kanamycin and cultured at 37 °C and 200 rpm for 24 h. Plasmid DNA was extracted, and the sgRNA sequences were amplified by PCR prior to high-throughput sequencing and subjected to robust ranking aggregation to perform differential gene analysis.

4.9. In Vivo Screen Assay

To screen for virulence factor genes related to neonatal sepsis, we applied CRISPRi screening in a pregnant rat model. Pregnant animals at E16 were anesthetized, and an 60 μL IPTG-induced or uninduced CRISPRi library ST95 ExPEC (10⁸ CFUs) suspension was inoculated into the vagina. At E18, the animals were euthanized by continuous exposure to high-concentration CO₂. Lavage fluid from the vagina, placenta, umbilical cord, fetal blood, and amniotic fluid samples was collected, homogenized under sterile conditions, transferred to liquid LB with kanamycin, and cultured at 37 °C and 200 rpm overnight. Then, the plasmid DNA was extracted and the sgRNA sequences were PCR-amplified and subjected to robust ranking aggregation to perform differential gene analysis.

4.10. Animal Model

To validate the role of the virulence factor genes in the infection process, pregnant SD rats were used in the in vivo study. For each group, three animals were included. Pregnant animals at E16 were anesthetized and 10⁸ CFUs of bacteria in 60 μL DPBS were inoculated into the vagina. At E20, euthanasia was performed by continuous exposure to high-concentration CO₂. The lavage fluid from the vagina, placenta, umbilical cord, fetal blood, and amniotic fluid samples were collected, homogenized under sterile conditions, and plated on LB plates. Some placentas and umbilical cords were fixed with formalin, embedded in paraffin, and stained with hematoxylin and eosin (HE).

4.11. Antibody

The primary antibody for ZO-1 was Rabbit mAb #8193 and for anti-occludin was Cell Signal Technology (Danvers, MA, USA) Rabbit mAb #91131; the anti-vimentin (D21H3) antibody was from Cell Signal Technology Rabbit (Danvers, MA, USA) mAb #5741, the anti-SFPQ antibody [EPR11847] was from Abcam (Waltham, MA, USA) ab177149, the anti-histone H3 was from Abcam (Waltham, MA, USA) ab1791, and the anti-β-actin antibody 20536-1-AP was from Proteintech (Chicago, IL, USA). A horseradish peroxidase (HRP)-conjugated secondary antibody (1:1000, ZSGB-Bio, Beijing, China, ZB-2301) was also used.

4.12. Statistical Analysis

GraphPad Prism version 10.4.1 (GraphPad Software, San Diego, CA, USA) was used to analyze the acquired data and to create graphs. An unpaired t-test was applied in the analysis of the penetration, internalization, and WB results. For comparison of more than two groups, one-way ANOVA was applied. A Gaussian distribution and the same-population SD were assumed. The threshold for statistical significance was set at p < 0.05. * stands for p < 0.05, ** stands for p < 0.01, **** stands for p < 0.0001.

5. Conclusions

We present a mechanistic model wherein the neonatal sepsis pathogen ST95 ExPEC employs a suite of virulence factors, with ibeA playing a pivotal role in driving trophoblast internalization to overcome the placental barrier. In response, the host deploys ASPHD1 as a target defense mechanism. These findings significantly advance our understanding of neonatal sepsis pathogenesis and illuminate the dynamic interplay between bacterial invasion and placental defense at the molecular level (Figure 6G).

Supplementary information: Other phenomenon found in the experiments: Syncytial trophoblasts have been proposed to exhibit high levels of autophagic activity and resistance to microbial infection [31]. Rapamycin, a well-known inducer of autophagy, was tested in vitro. Syncytial trophoblasts treated with rapamycin were a more effective barrier against ST95 ExPEC than the control syncytial trophoblasts (Supplementary Figure S1G), suggesting that rapamycin could be a potential therapeutic agent against ST95 ExPEC infection. However, p62 levels did not change after rapamycin plus ST95 ExPEC treatment, and the autophagy inhibitors 3-MA and bafilomycin did not affect ST95 ExPEC penetration into syncytial trophoblasts (Supplementary Figure S1H). This enhanced barrier effect may result from inhibition of mTOR. A recent study reported rapamycin could induce BeWo to a syncytial state [32], which may explain this phenomenon. The levels of the tight junction protein occludin decreased after ST95 ExPEC infection. LPS was reported to induce inflammatory cytokine secretion and downregulate tight junction protein expression in the blood–brain barrier, which may explain the tight junction protein reduction in BeWo cells upon ST95 ExPEC stimulation [33].

Abbreviations

The following abbreviations are used in this manuscript:

CRISPR	Clustered Regularly Interspaced Short Palindromic Repeats
E. coli	Escherichia coli
SD pregnant rat	Sprague Dawley Pregnant Rat
LB	Lysogeny Broth
sgRNA	Single-Guide RNA
ST95 ExPEC	Multilocus Sequence Type 95 (ST95) Extraintestinal Pathogenic Escherichia coli (ExPEC)
IPTG	Isopropyl β-D-1-thiogalactopyranoside
FBS	Fetal Bovine Serum
F-12K	Kaighn’s Modification of Ham’s F-12 Medium
CFU	Colony Forming Units
mTOR	Mammalian Target of Rapamycin
LDH	Lactate Dehydrogenase
CAPE	Caffeic Acid Phenethyl Ester
3-MA	3-Methyladenine
PCA	Principle Component Analysis
MOI	Multiplicity Of Infection
FAD	Flavin Adenosine Dinucleotide
LPS	Lipopolysaccharide

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/ijms27041661/s1.

Author Contributions

Conceptualization, X.C. and Y.W.; methodology, X.C. and Y.W.; software, X.C.; validation, X.C.; formal analysis, X.C.; investigation, X.C., X.L., P.Z. and R.X.; resources, Y.W.; data curation, X.L.; writing—original draft preparation, X.C.; writing—review and editing, X.L. and Y.W.; visualization, X.C.; supervision, Y.W.; project administration, Y.W.; funding acquisition, X.C. and Y.W. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki. The acquisition of clinical ST95 ExPEC strains (Approval No. 2019-k-350, Approval date 1 October 2019) was reviewed and approved by the Institutional Review Board (IRB) of Capital Medical University Affiliated Children’s Hospital, Beijing, China. The animal study (Approval No. DWLL2025017, Approval date 10 July 2025) was reviewed and approved by the Ethics Committee of the Children’s Hospital, Capital Institute of Pediatrics, Beijing, China. All methods were performed by relevant guidelines and regulation.

Informed Consent Statement

Informed consent was obtained from the parent of the neonates (who were minors) in written form prior to sample collection. For minors, assent was also obtained from the parents as legal guardians. The consent process included detailed explanation of the study purpose, procedures, and usage of samples for research. The study adhered to ethical standards for research involving human participants.

Data Availability Statement

Data is contained within the article or Supplementary Material.

Conflicts of Interest

The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Funding Statement

This work was supported by the Natural Science Foundation of Beijing Municipality [Grant Nos. 7232009, 7244289], the National Natural Science Foundation of China [Grant No. 82574169], the Cross-cooperation project of the Beijing Science and Technology New Star Program [Grant No. 20240484724], the High Level Public Health Technical Personnel Construction Project [Subject leaders-03-02], the Research Foundation of Capital Institute of Pediatrics [YCYJ-2025-08], and the Beijing Chaoyang District Postdoctoral Research Foundation 2024. The funders had no role in the study design, data collection and analysis, decision to publish, or manuscript preparation.