CD14 and ALPK1 Affect Expression of Tight Junction Components and Proinflammatory Mediators upon Bacterial Stimulation in a Colonic 3D Organoid Model

Cd14 and Alpk1 both encode pathogen recognition receptors and are known candidate genes for affecting severity in inflammatory bowel diseases. CD14 acts as a coreceptor for bacterial lipopolysaccharide (LPS), while ALPK1 senses ADP-D-glycero-beta-D-manno-heptose, a metabolic intermediate of LPS biosynthesis. Intestinal barrier integrity can be influenced by CD14, whereas to date, the role of ALPK1 in maintaining barrier function remains unknown. We used colon-derived 3D organoids, first characterised for growth, proliferation, stem cell markers, and expression of tight junction (TJ) components using qPCR and immunohistochemistry. They showed characteristic crypt stem cells, apical shedding of dead cells, and TJ formation. Afterwards, organoids of different genotypes (WT, Il10−/−, Cd14−/−, and Alpk1−/−) were then stimulated with either LPS or Escherichia coli Nissle 1917 (EcN). Gene expression and protein levels of cytokines and TJ components were analysed. WT organoids increased expression of Tnfα and tight junction components. Cd14−/− organoids expressed significantly less Tnfα and Ocln after LPS stimulation than WT organoids but reacted similarly to WT organoids after EcN stimulation. In contrast, compared to WT, Alpk1−/− organoids showed decreased expression of different TJ and cytokine genes in response to EcN but not LPS. However, Western blotting revealed an effect of ALPK1 on TJ protein levels. These findings demonstrate that Cd14, but not Alpk1, alters the response to LPS stimulation in colonic epithelial cells, whereas Alpk1 is involved in the response upon bacterial challenge.


Introduction
Inflammatory bowel disease (IBD) onset involves the interplay between microbiota, environmental conditions, genetic factors, and a disrupted intestinal barrier [1]. Several mouse models of IBD have been established [2] including the wellstudied Il10 (interleukin-10) deficiency model. This model is characterised by a dysregulated immune response to enteric microflora leading to the onset of colitis through a disrupted barrier due to increased levels of interferon gamma and tumor necrosis factor alpha (TNFα) [3]. Using the Il10 -/mouse model and quantitative trait locus (QTL) analyses, several studies have identified genetic loci associated with susceptibility to IBD. These studies revealed ten cytokine deficiency-induced colitis susceptibility (Cdcs) loci and candidate genes with a potential influence on colitis onset [4][5][6][7][8][9]. One of these genes is Cd14 (Cluster of differentiation 14) in the Cdcs6 locus located on chromosome 18 [7]. CD14 acts as a coreceptor of Toll-like receptor (TLR) 4 and is directly involved in the detection of lipopolysaccharide (LPS) and activation of NF-κB [10]. In previous studies, our group demonstrated that CD14 has a protective function on barrier integrity in a dextran sodium sulfate-(DSS-) induced acute colitis mouse model [11]. We also showed that in wild-type (WT) but not in Cd14 -/mice, monoassociation with Escherichia coli Nissle 1917 (EcN) leads to increased expression of tight junction (TJ) components, associated with protection against bacterial translocation [12].
Another candidate gene revealed by these QTL studies is Alpk1 (alpha-protein kinase 1) which is located on chromosome 3 in the Cdcs1 locus [4,5]. Knockout of the Alpk1 gene in mice was recently shown to lead to severe colitis when infected with Helicobacter hepaticus [13]. ALPK1 functions as a pattern recognition receptor for adenosine diphosphateheptose (ADP-Hep), a precursor of LPS [14]. In human epithelial cells, ADP-Hep of Gram-negative bacteria activates NF-κB-dependent inflammatory interleukin-8 (IL-8) secretion via an ALPK1-TIFA-TRAF6 axis [15]. As mice lack a functional Il-8 gene, the chemokines CXCL1, CXCL2, and CXCL5 are regarded as functional homologues [16].
To study the specific impact of Il10, Cd14, and Alpk1 on intestinal epithelial cell (IEC) response to bacterial stimulation, we used colonic organoid cultures derived from isolated intestinal stem cells (ISCs) [17]. These ISCs can differentiate into all colonic epithelial lineages including colonocytes, goblet cells, and several enteroendocrine cell types [18]. In the present study, colonic organoids of different knockout mouse strains were stimulated with LPS or EcN and the expression of proinflammatory mediators and tight junction components was determined. Here, we report that IEC Cd14 and Alpk1 both affect cytokine and TJ component expression upon bacterial challenge.

Organoid Kinetics.
Matrigel® was dissolved using icecold DPBS, and crypts were separated by drawing the suspension up once through a syringe using a 27G 1/2 ″ needle. Cells were washed and seeded into fresh Matrigel®, which was allowed to solidify for 30 minutes at 37°C, and afterwards overlaid with 500 μl organoid growth media. Pictures were taken on days 1, 3,5,7,9,11, and 13 after seeding. At each time point, three samples were obtained for RNA isolation by pooling two wells of a 24-well plate for each sample.
2.6. TUNEL Staining. Preparation of Matrigel®-embedded organoids was performed as per immunofluorescence staining. TUNEL staining was performed using an In Situ Cell Death Detection Kit, Fluorescein (Roche, Basel, Switzerland), according to the manufacturer's instructions.

Stimulation with LPS.
As we wanted to activate CD14dependent signalling, a low dose of LPS was necessary as it is known that higher concentrations of LPS result in CD14independent TLR4 signalling [19,20]. In previous experiments using the mouse epithelial cell line CMT93, we tested different LPS concentrations and time points [11]. In our 2 Stem Cells International hands, 0.1 μg/ml and 6 hours were the best concentration and incubation time to measure CD14-dependent cytokine induction. After 10 days of growth, colonic organoids were stimulated with LPS (Sigma-Aldrich, St. Louis, USA) as described previously [11]. In brief, growth media was removed and replaced with 500 μl fresh growth media without Cellshield and containing either 0.1 μg/ml LPS or no LPS as the control. After 6 hours of incubation at 37°C and 5% CO 2 , media was removed and the Matrigel® was dissolved in ice-cold DPBS. Two to 6 wells were pooled, and cells were centrifuged, resuspended in RNA Quick-RNA™ Micro Prep Kit Lysis Buffer (Zymo Research, Irvine, USA), and stored at -80°C for later RNA isolation. KG). Proteins were extracted and separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis, transferred to a nitrocellulose membrane, and immunoblotted using primary antibodies against occludin, claudin8, zonula occludens-1 (Invitrogen), and claudin4 (Abcam). Immunoblotting for GAPDH (GenScript USA Inc., NJ, USA) was used as an internal control. After incubation with the secondary antibody (donkey anti-rabbit IgG (HRP); Abcam), the membrane was developed with a chemiluminescence solution (Bio-Rad Laboratories, Hercules, California, USA) and images were acquired using a ChemiDoc™ Touch Imaging System (Bio-Rad). Blot analysis was performed via ImageJ software (open source).

Statistical Analysis.
All statistical analyses were performed using GraphPad Prism6® software (San Diego, USA). Values represent the means with 95% CI or SEM. All values were normalized to the corresponding unstimulated controls in each independent experiment. Accordingly, unstimulated control values were calculated to a value of 1. When two conditions were compared, unpaired t-tests were performed. In other cases, one-way analysis of variances followed by Dunnett's multiple comparison tests were used to compare values of stimulated samples of all three knockout genotypes to WT organoids. In the case of unequal variances, a Brown-Forsythe and Welch ANOVA test was performed. P < 0:05 was considered significant. * indicates P < 0:05, * * indicates P < 0:01, * * * indicates P < 0:001, and * * * * indicates P < 0:0001.

Results
day 5 and remained stable until day 11, after which they rapidly decreased until day 13.
To characterise cell proliferation within organoids, Ki67 expression was determined on the mRNA level by qPCR. An increase in Ki67 gene expression occurred up to day 5 with a drop in expression at day 7. After a media change on day 7, Ki67 expression increased and kept increasing until another media change on day 10, after which expression became quite variable. This resulted in widely distributed expression values on day 11. On day 13, Ki67 mRNA was no longer detectable in any organoids. Gene expression of the IEC genes Smoc2 and LgR5 was analysed and showed a similar expression pattern as Ki67. In contrast, gene expression of the proliferation inhibitor and epithelial differentiation marker Clca4b [21] was low and mostly stable up to day 9, after which expression increased until day 13. Furthermore, gene expression of TJ components such as Ocln, Tjp1, Cldn4, and Cldn8 was analysed. Similar expression levels were detectable between day 1 and day 13 for Ocln and Tjp1, while Cldn4 increased and Cldn8 decreased rapidly on day 13 (Figure 1(d)). Thus, we decided to perform all further experiments using organoids after 10 days of growth, where the morphology showed crypts and the proliferation level was still high.

Colonic Organoids Represent Fully Differentiated Colon
Epithelium. We performed H&E staining on 10-day-old organoid sections to analyse the differentiation and growth status of the IECs (Figure 2(a)). Around the circular lumen, a fully differentiated epithelium with crypts formed, consisting primarily of enterocytes and mucus-containing goblet cells. A few LGR5-positive cells (Figure 2(b)) were present at the base of crypts, whereas high amounts of Ki67positive cells (Figure 2(c)) were located at the base and sides of crypt and were not found elsewhere in the organoids. Dead cells collected in the lumen as confirmed by whole mount TUNEL staining (Figure 2(d)). The organoid cell layer stained positive for epithelial cell adhesion molecule (EPCAM or CD326) which confirmed these cells as epithelial cells (Figure 2(e)). These cells were also positive for the intracellular TJ proteins occludin (OCLN) (Figure 2(f)), tight junction protein 1 (TJP1) (Figure 2(g)), and claudin-8 (CLDN8) (Figure 2(h)).

Decreased Expression of Tight Junction Components and
Proinflammatory Response in Cd14 -/-Epithelial Cells upon LPS Stimulation. After 6 hours of LPS stimulation, RNA or protein samples were taken from the organoids and expression levels of different genes involved in tight junctions (Ocln, Tjp1, Cldn4, and Cldn8) (Figure 3 and Supplemental Figure 1) and immune response (Tnfα, Cxcl1, Cxcl2, and Cxcl5) were analysed (Figure 4). To adjust batch effects, all values were normalized to the corresponding unstimulated controls in each independent experiment. Compared to WT organoids, Ocln expression significantly increased in Il10 -/organoids after exposure to LPS but was significantly  Stem Cells International decreased in Cd14 -/organoids after LPS stimulation (Figure 3(a)). Western blot analysis of OCLN showed no differences in all organoids after 6 h of LPS stimulation (Figure 3(b)). Gene and protein expression of Tjp1 was not altered in any of the organoids after LPS challenge (Figures 3(a) and 3(b)). LPS exposure significantly increased Cldn4 expression in WT and Il10 -/and slightly in Alpk1 -/organoids (Figure 3(a)). In Cd14 -/organoids, Cldn4 expression decreased slightly upon LPS exposure. These differences were not detectable on protein level (Figure 3(b)). As for Cldn8, expression levels remained unchanged for all genotypes (Figure 3(a)); however, CLDN8 expression was increased in Alpk1 -/organoids after LPS stimulation (Figure 3(b)). Upon LPS stimulation, gene expression of proinflammatory Tnfα was increased in all organoids, but this increase was significantly less in Cd14 -/organoids compared to WT organoids (Figure 4(a)). Protein levels of TNFa proteins also increased in all organoids especially in Il10 -/epithelial cells. In contrast, TNFα secretion was not increased in Cd14 -/epithelial cells after LPS stimulation (Figure 4(a)). Chemokines Cxcl1, Cxcl2, and Cxcl5 were increased after LPS stimulation in both investigated organoid genotypes, WT and Alpk1 -/- (Figures 4(b)-4(d)). However, multiplex analysis of CXCL1, CXCL2, and CXCL5 revealed only an increase in WT organoids. Alpk1 -/epithelial cells showed only a modest increase in chemokine expression. Thus, LPS triggers a proinflammatory effect in all observed organoids but had lower effects in Cd14 -/and Alpk1 -/epithelial cells. Additionally, expression of TJ components Ocln and Cldn4 is affected by LPS in Cd14 -/organoids and CLDN8 seems to be altered in Alpk1 -/epithelial cells.

Alpk1 Deficiency Affects Tight Junction Component Expression after E. coli Nissle 1917 Infection of Colonic
Organoids. After 6 hours of EcN stimulation, gene and protein expression of tight junctions and inflammatory cytokines was measured ( Figures 5 and 6 and Supplemental Figure 1). WT and Il10 -/organoids showed no changes in Ocln and Tjp1 expression compared to uninfected controls (Figures 5(a) and 5(b)); Ocln expression was significantly increased in Cd14 -/organoids ( Figure 5(a)). Upon EcN infection, Alpk1 -/organoids had slight but not statistically significant decreased Ocln expression ( Figure 5(a)). However, Western blot analysis showed an increased OCLN level compared to WT organoids ( Figure 5(b)).
No difference in chemokine secretion between Alpk1 -/organoids and WT epithelial cells was observed in multiplex analysis of supernatants (Figures 6(b)-6(d)).
Together, our data show (i) decreased Ocln and Tnfα gene expression in LPS-stimulated Cd14 -/organoids compared to WT epithelial cells; (ii) increased CLDN8 protein expression after LPS stimulation in Alpk1 -/cells; (iii) decreased Cldn8, Cxcl1, and Cxcl5 gene expression and increased protein expression of TJ components such as CLDN4, CLDN8, and OCLN in EcN-stimulated Alpk1 -/organoids compared to stimulated WT organoids; and (iv) less Tnfα/TNFα production in Cd14 -/cells compared to WT epithelial cells. In conclusion, these data indicate that Cd14, but not Alpk1, alters the response to LPS stimulation in colonic epithelial cells, whereas Alpk1 is more involved in the proinflammatory response to bacterial challenge.

Discussion
In the present study, we utilized colon-derived 3D organoids derived from WT, Il10 -/-, Cd14 -/-, and Alpk1 -/knockout mice to investigate the response to either purified LPS or whole Gram-negative bacteria. We show for the first time a role for ALPK1 in TJ component regulation after infection with the Gram-negative bacteria EcN. In addition, our results corroborate the role of CD14 in LPS-triggered proinflammatory and TJ component responses [11]. Although there are reports of Il10 being expressed in IECs [22], IL10 itself is primarily produced by immune cells where it functions to regulate the intestinal homeostasis and barrier integrity via its anti-inflammatory effects [23,24]. In our study, the absence of IL10 did not alter the response of colonic organoids to EcN compared to WT cells.
There are several variations described to culture, separate, and passage 3D organoids which result in different periods of growth and differentiation [18,25,26]. Our isolated crypt stem cells developed into colonospheres within 24 hours. The observed growth correlated with an increase in RNA concentration, proliferation, and expression of stem cell genes in the first 5 days. Organoid growth and differentiation are highly dependent on the supplied media  LPS: Densitometric quantification normalized to GAPDH was calculated as change from the unstimulated control. RD = relative density %. Stimulated versus unstimulated values were compared using unpaired t-tests. Welch's correction was used in case of unequal variances. One-way analysis of variance followed by Dunnett's multiple comparisons was performed. Each knockout genotype was compared to WT. * * P < 0:01, * * * P < 0:001.

6
Stem Cells International components. We assume that the drop in Ki67, Lgr5, and Smoc2 expression after day 5 resulted from consumption of the differentiation inhibitor Y-27632. After a media exchange, organoids proliferated once more and formed more crypts. Eleven days after seeding, organoids began to break open, and by day 13, they were almost uniformly disrupted. In another study by Thalheim et al. [27] investigating the growth pattern of intestinal organoids, an equilibrium in the different epithelial cell type fractions was reached after 10 days. When we observed our organoids after 10 days, LGR5- were compared using unpaired t-tests. In the case of unequal variances, Brown-Forsythe and Welch ANOVA test was performed. In all other cases, a one-way analysis of variance followed by Dunnett's multiple comparisons was performed. Each knockout genotype was compared to WT. * P < 0:05, * * P < 0:01, * * * P < 0:001, and * * * * P < 0:0001. 7 Stem Cells International and KI67-positive IECs were at the base of crypts indicating proliferating stem cell populations. Both enterocyte and mucus-producing goblet cell population were observed by H&E staining, immunofluorescence, and cellular junction staining, and intestinal desquamation and cell turnover were visible. Therefore, we chose day 10 to begin our stimulation and infection experiments since at this time point colonic organoids were viable, differentiated, and demonstrated mature cellular junctions and crypt-villus structure.
IEC is primarily exposed to bacteria and LPS on their apical surface; however, the basolateral surface is also exposed when the barrier is compromised or upon infection with invasive pathogens. Several studies have observed an effect of physiological concentrations of basally and apically applied LPS on IEC TJ permeability [28][29][30]. The way of infection in 3D organoids is dependent on the regarded strain [31]. In a study of Sabharwal et al. in 2016, polarized T84 cells were infected with EcN [32]. Apical EcN infection triggered only a slight upregulation of Cxcl1 and an additional downregulation of Il8 expression. When infecting from the basolateral side, EcN greatly enhanced expression of Il8 and Cxcl1 [32]. In our organoid system, stimulation and infection is performed on the basolateral surface as EcN infection triggers epithelial activation at this site.
LPS is a pathogen-associated molecular pattern that is recognized by TLR4/CD14. Signalling via TLR4/CD14 leads to NF-κB activation and subsequent proinflammatory gene expression [33]. In our study, we used a LPS concentration of 0.1 μg/ml, which was previously shown to induce an epi-thelial response in vitro [11]. After 6 hours of LPS stimulation or EcN infection, all organoid lines (WT, Cd14 -/-, Il10 -/-, and Alpk1 -/-) had increased expression of Tnfα, proving a positive reaction to the stimulus. However, the increase in Tnfα/TNFα expression in the Cd14 -/organoids was lower compared to WT organoids, validating the role of CD14 in LPS-triggered proinflammatory signalling [20]. ALPK1 does not recognize LPS itself but rather a metabolic intermediate involved in LPS biosynthesis. Accordingly, our results confirmed that the lack of ALPK1 barely affects IEC proinflammatory response to LPS exposure alone, as TNFα expression was similar compared to LPS-stimulated WT cells. Upon bacterial infection of IECs, ALPK1 has been shown to signal through TRIF/TRAF6 to NF-κB, triggering secretion of proinflammatory cytokines like IL8 and TNFα [34]. However, upregulation of chemokine Cxcl1 and Cxcl5 was significantly less in Alpk1 -/organoids infected with EcN compared to WT organoids, but an Alpk1 knockout did not result in decreased levels of proinflammatory mediators in the supernatants upon bacterial stimulation of colonic organoids. Therefore, in our experimental setup, the lack of ALPK1 did not result in decreased levels of proinflammatory mediators suggesting a minor role of the ALPK1 signalling pathway in epithelial cells.
Upon LPS stimulation of WT organoids, we did not observe changes in the expression of TJ component genes Ocln, Tjp1, or Cldn8 but did record a significant increase in Cldn4 expression. EcN infection of WT organoids increased Cldn4 and Cldn8 expression, suggesting a positive effect of   Figure 5: Expression of tight junction genes after E. coli Nissle stimulation. Wild-type (WT) and knockout organoids were stimulated with E. coli Nissle 1917 for 6 hours and analysed for changes in tight junction component gene and protein expression. Differences between knockout and WT organoids were investigated. Values were normalized to unstimulated controls within each genotype for each experiment and represent the means with 95% CI. Unstimulated: WT n = 3-30, Cd14 -/n = 3-11, Il10 -/n = 3-9, and Alpk1 -/n = 3-9. Stimulated: WT n = 6 -29, Cd14 -/n = 3-12, Il10 -/n = 3-8, and Alpk1 -/n = 3-9. (a) Gene expression. RQ = relative quantification. (b) Detection of TJ proteins by Western blot analysis of lysates from organoids after 6 h of EcN stimulation. Densitometric quantification normalized to GAPDH was calculated as change from the unstimulated control. RD = relative density %. Stimulated versus unstimulated values were compared using unpaired t-tests. Welch's correction was used in case of unequal variances. One-way analysis of variance followed by Dunnett's multiple comparisons was performed. Each knockout genotype was compared to WT. * P < 0:05, * * P < 0:01, * * * P < 0:001, and * * * * P < 0:0001.
In addition, infection with whole Gram-negative bacteria revealed a significant increase in all analysed TJ proteins in Alpk1 -/organoids, while RNA expression levels of Tjp1 and Cldn8 were decreased. These results indicate that ALPK1 plays an important role regulating tight junction component expression in IECs which could, in turn, lead to an altered IEC barrier. In contrast, ALPK1 did not affect the nonhematopoietic compartment in a H. hepaticus-induced colitis mouse model. Rather, ALPK1 influenced the regulation of IL12 production in bone marrow-derived macrophages [13].
Our group previously described Alpk1 as a Cdcs1 locus candidate gene for IL10 deficiency-induced colitis severity in mice [4] and recently excluded a T cell-dependent role for ALPK1 in colitis progression [5]. Our present study shows an influence of Alpk1 on epithelial barrier components in response to EcN and therefore perhaps to tolerance to commensal bacteria. A defect in Alpk1 could therefore contribute to a disruption of barrier function upon bacterial infection and be a cause for exacerbation of colitogenic inflammation. Future studies concerning IBD and intestinal barrier integrity should therefore investigate whether Alpk1dependent signalling is involved in disease progression.

Data Availability
The raw data supporting the conclusions of this manuscript are available from the corresponding author on reasonable request to any qualified researcher.