Lysophosphatidic acid and sphingosine-1-phosphate are apical polarity cues in multiple organoid systems

SUMMARY

Apicobasal polarization is crucial for tissue organization during in vivo development and in human organoid models. Extracellular matrix (ECM) signaling typically provides a basal cue, and intestinal and lung organoids reverse polarity from apical-in to apical-out after ECM removal. However, ECM-free brain organoids maintain apical-in polarity, suggesting that media components may influence polarity. Exposing brain organoids to serum induced apical-out orientation. Lysophosphatidic acid (LPA), present in the medium of prior apical-out techniques, was identified as the causative factor. LPA-induced apical-out orientation in brain organoids occurred within 1 day, lasted at least 1 month, and was optimal at human cerebrospinal fluid LPA concentrations. Sphingosine-1-phosphate (S1P) induced similar apical-out polarization. Pharmacological studies revealed that LPA/S1P act via a G-protein coupled receptor/RhoA pathway. Finally, LPA induced apical-out polarity in patient-derived human lung and intestinal organoids, iPSC spheres, and multilineage iPSC-derived intestinal organoids. These findings indicate that LPA signaling is a critical apical polarity cue in multiple tissues.

In brief

Tidball et al. reveal that LPA and S1P induce apical-out epithelial orientation in brain, lung, and intestinal organoids through GPCR/RhoA signaling. These phospholipids, present in biological fluids and many culture medias, act as apical cues, highlighting their pivotal role in organoid and tissue organization.

INTRODUCTION

Apicobasal polarity is crucial in epithelial development.¹ Many epithelial cell containing organoids use extracellular matrix (ECM) protein hydrogels to achieve a 3D structure.^2–5 Integrin signaling from the ECM provides a basal polarity cue that allows these structures to form primarily into apical-in sphere-like cysts.^{3 ,4} Removal of intestinal, lung, and retinal organoids from ECM hydrogels causes polarity inversion with an apical-out orientation.^3,5,6 Apical-out organoids are important for understanding apical-specific functions such as nutrient absorption, microbiome-host interaction, air-tissue interface, cilia motility, and apically located ligand-receptor cell signaling.^3,5,7,8 Apical-out orientation has also been induced in intestinal organoids using an antibody that blocks the function of beta1 integrin³ and in mammary epithelial cysts beta1-integrin knockout (KO).⁴ While these results highlight the importance of integrin signaling as an extrinsic basal signaling cue, none of these studies explain why removing a basal cue would cause re-orientation of the polarity, rather than maintaining the current polarity. Therefore, we hypothesized that, while ECM serves as a basal polarity cue, organoids may also be responding to an apical cue present in the culture media.

To test this hypothesis, we sought to make apical-out human induced pluripotent stem cell (hiPSC)-derived brain organoids by generating ECM-free spheres aggregated from dissociated neuroepithelial cells. Standard media formed typical apical-in brain. However, with the addition of 1% fetal bovine serum, all organoids became apical-out, suggesting that an extrinsic apical cue present in serum induces apical-out organoid formation. Here, we show clear evidence that phospholipids found in serum, lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P), cause apical-out formation through activation of G-protein coupled receptors (GPCRs) and RhoA. We further demonstrate that LPA-dependent apical-out orientation occurs at physiological LPA concentrations found in cerebrospinal fluid (CSF) and apical-out orientation occurs in direct response to CSF exposure. Finally, LPA caused apical-out orientation in patient-derived lung, intestinal organoids, and undifferentiated hiPSC spheres. Both LPA and S1P caused an apical-out orientation in human iPSC-derived intestinal organoids through the same GPCR/RhoA pathway seen in brain organoids, indicating that these phospholipids are widespread apical cues across tissue types and developmental time points.

RESULTS

LPA causes apical-out polarity in ECM-free brain organoids

To investigate a hypothetical apical cue, we adapted our SOSR-CO protocol to create ECM-free brain organoids by re-aggregating hPSC-derived neuroepithelial cell suspensions into spheres.⁹ Like other brain organoid models, these neuroepithelial spheres developed multiple rosettes with apical-in lumens within 48 h, as shown by apical markers filamentous actin (F-actin) and ZO1 (Figures 1A and 1B). While all current published techniques generate apical-in brain organoids, some protocols clearly display a transitory apical-out state when KO serum replacement (KOSR) is present in the media.^10–12 Furthermore, experiments that showed cell-line-based cysts formed apical-out after ECM or beta1-integrin removal all used serum-containing culture media, unlike brain organoid media that is serum free.^4,13 Therefore, we suspected that serum and KOSR may contain an apical cue. Adding 1% serum to the culture media caused robust apical-out orientation in every organoid with ZO1 and F-actin on the outside edge (Figures 1C and S1A). These organoids continued to express neural tube markers NESTIN, PAX6, N-cadherin, and the forebrain marker FOXG1 (Figures S1B–S1D). Dividing cells labeled with phosphorylated vimentin (S55) or TPX2 were almost exclusively on the outside edge indicating typical interkinetic nuclear migration (Figures S1E and S1F).

Figure 1. LPA induces apical-out orientation in brain organoids.

(A) Timeline of the adapted SOSR-CO protocol, highlighting the generation of single cell suspensions at day 4 and aggregation using Aggrewell.

(B–E) Confocal images of cryosectioned organoids after 48 h of exposure to test compounds and immunolabeling for the apical markers ZO1 and F-actin (DNA counterstained with Bisbenzimide).

(F–I and K–M) Whole-mount confocal maximum projections (bottom 100 μm) of organoids imaged for ZO1-EGFP to label apical tight junctions.

(J) Quantification of the percent apical-out in individual organoids based on ZO1-EGFP and immunostaining from 4 different pluripotent stem cell lines. ****p < 0.001 by Kruskal-Wallis test (J). Each dot is an organoid. Data are represented as mean ± SD. n = 31 (vehicle) and 35 (LPA) organoids across the 4 different lines with one experiment for each.

Organoids are day 7 of differentiation. Scale bars, 100 μm.

We next sought to identify the factor in serum that induces apical-out orientation. In addition to serum, KOSR or advanced DMEM/F12-containing media are also used in published apical-out organoid protocols.^3,5,11,12 Both products contain AlbuMAX, a lipid-rich serum albumin product. To test whether lipids may be involved in the apical-out orientation, we boiled the serum at 98°C for 10 min to denature proteins. This failed to block the apical-out serum effect (Figure 1D), suggesting that a lipid may be the apical cue. Previous 2D studies found LPA caused enlarged neural rosette lumens to form.¹⁴ Therefore, we tested the effect of LPA on ECM-free brain organoids and found that 200 nM was sufficient to induce the apical-out orientation (Figure 1E). This concentration is very close to the 190 ± 80 nM concentration found in human CSF.¹⁵ a Total LPA concentration from Albumax I in 20% KOSR-containing media has been measured at approximately 170 nM.¹⁶ Advanced DMEM/F12 contains 400 mg/L of Albumax II, which also contains LPA, but at an unpublished concentration. To confirm our hypothesis that these medias resulted in the apical-out orientation observed in previous studies, we replaced DMEM/F12 with advanced DMEM/F12 or added 20% KOSR. At this point, we began using a commercially available iPSC line with an EGFP fused to one allele of the endogenous ZO1 gene (TJP1) to quickly indicate the position of the apical domain. In each case, the organoids were apical out, similar to the 1% serum and 200 nM LPA (Figures 1F–1I). We confirmed LPA-dependent apical-out phenotypes across four pluripotent stem cell lines including one human embryonic stem cell and three iPSCs. Two lines were female, two were male, three were obtained from outside vendors, while one iPSC line was reprogrammed and validated in previous publication.¹⁷ None of the vehicle-treated organoids were apical out while all 100 nM LPA-treated organoids were greater than 62% apical out by area with an average of 93 ± 10% (Figure 1J). To investigate the effect of LPA when ECM is present, we added the spheres to wells coated with 100% Geltrex with 0, 100, or 1,000 nM LPA. In each instance, when Geltrex was present (Figures 1K–1M), the organoids formed apical in with larger lumens than without ECM (Figure 1F), showing that the effects of the ECM on orientation are greater than LPA.

A published spinal cord organoid system found apical-out orientation was dependent on fibroblast growth factor 2 (FGF2) addition to the media.¹⁸ We tested whether adding FGF mimics or alters LPA-dependent polarization. At concentrations of up to 2 ng/mL of FGF2, we saw no apparent effect on polarity with or without LPA. At 20 ng/mL, FGF2 still did not impact polarity but decreased the homogeneity of the ZO1-labelled tight junction barrier (Figure S2).

To characterize these apical-out brain organoids, we immunostained for several important apical and structural markers. Apical markers, PALS1, ZO1, and cingulin transitioned from internal lumens to the external surface with LPA exposure (Figures 2A–2J). Myosin-IIb also became localized to the apical surface and colocalized with F-actin (Figures 2K–2O). Acetylated-tubulin showed a dramatic change from disorganized radial orientation around apical-in lumens in vehicle-treated organoids (Figure 2K) to forming microtubule networks perpendicular to the apical membrane (Figures 2M–2O). High-magnificonfocal microscopy showed strong co-localization of F-actin with both ZO1 and MyoIIB (Figures 2H–2J and 2M–2O). The important apical protein, PAR3, localizes to apical domains and ZO1-EGFP-labeled tight junctions (Figures 2P and 2T). Adherens junction marker proteins N-cadherin and β-catenin appear immediately basal to the tight junctions (Figures 2Q, 2R, 2U, and 2V), while Arl13B-labeled primary cilia project above the apical surface (Figures 2S and 2W). All of these proteins in the LPA-treated organoids are in the correct location for a mature apicobasally polarized cortical neuroepithelium with an outward-facing apical membrane.

Figure 2. Structural and apical marker characterization in organoids.

Neuroepithelial cell spheres were treated with LPA (1,000 nM B–E; 100 nM G–J, L–W) or vehicle (A, F, and K). Whole-mount confocal imaging was performed with 20× objective with a total z distance of 100 μm (A, B, F, G, K, and L) or 60× objective over 10 μm (C–E, H–J, M, and O). Images are maximum z-projections (A–C, F–H, K–M, and P–S) or orthogonal sections (D, E, I, J, N, O, and T–W) with the apical side up and basal side down. (A–E) are from a control iPSC WT line generated from foreskin fibroblasts.¹⁷ (F–O) are from the commercially available WT iPSC control line RPChiPS8023G1.

(P–W) are from the commercially available iPSC line AICS-0023. Arrowheads highlight tight junctions. Organoids are day 7 of differentiation. Scale bars, 100 μm unless labeled otherwise.

LPA-dependent orientation is specific to LPA/S1P, is reversible, and is optimal at CSF concentrations

Many phospholipids are present in serum besides LPA.¹⁹ We sought to determine if any of these other phospholipids induce apical-out orientation. We tested 48-h exposure to the following phospholipids at 1 μM: cardiolipin, lysophosphatidylcholine, lysophosphatidylethanolamine, PA, phosphatidylethanolamine, phosphatidylinositol, phosphatidylserine, and S1P. Initial experiments (Figure 1) used LPA dissolved in DMSO; however, these other phospholipids required fatty acid-free BSA to be dissolved in aqueous solution. Therefore, LPA dissolved in DMSO was compared with BSA with no difference in the apical-out orientation (Figure S3). In the phospholipid screen, only PA and S1P induced apical-out orientation (Figures 3A–3J). While 100 nM LPA or S1P was sufficient to cause apical-out orientation, 1,000 nM but not 100 nM was effective for PA (Figures 3K, 3L, 3O, and 3P). From these data, we hypothesized that PA was being converted into LPA. This conversion is achieved by phospholipase A2 (PLA2) that cleaves one fatty acid chain (Figure 3S). To test this idea, we added the cell-permeable PLA2 inhibitor N-(p-amylcinnamoyl) anthranilic acid (ACA) to the organoid media. ACA (50 μM) partially blocked the apical-out orientation in PA-treated organoids but not in LPA-treated organoids (Figures 3M and 3Q). Interestingly, lisofylline (100 μM), an inhibitor of LPA acyltransferase (LPAAT)—the enzyme that converts LPA into PA—also partially blocked the PA-dependent phenotype (Figures 3N and 3R). This result fits with the published finding that LPAAT can work in the reverse direction, converting PA into LPA.²⁰ Therefore, PA seems to cause apical-out orientation through conversion to LPA. In contrast, S1P is an independent effector of apical-out polarity, as it cannot be converted into LPA (and vice versa).

Figure 3. Specificity of lipid-induced apical-out orientation.

(A–J) Spheres were treated for 48 h with BSA vehicle or 1 μM of each indicated phospholipid, including LPA, lysophosphatidylcholine (LPC), lysophosphatidylethanolamine (LPE), PA, phosphatidylethanolamine (PE), phosphatidylinositol (PI) phosphatdylserine (PS), and S1P.

(K, L, O, and P) Dose-response comparison between LPA and PA at 100 and 1,000 nM.

(M, N, Q, and R) 100 nM LPA (M and N) or 1000 nM PA (Q and R) were treated with either the PLA2 inhibitor (ACA, 50 μM) (M,Q) or LPAAT inhibitor (lisofylline, 100 μM) (N and R).

(S) A schematic of the PA to LPA conversion, enzymes, and inhibitors.

All images are maximum projections of ZO1-EGFP confocal imaging stacks for the bottom 100 μm of each sphere. Organoids are day 7 of differentiation. Scale bars, 100 μm.

We measured the relationship between LPA concentration and apical-out orientation. For 3 and 10 nM LPA, we found an intermediate phenotype with apical-out and apical-in regions (Figures 4A and 4B). For 30–300 nM, we saw nearly ideal apical-out ZO1-EGFP, while 1,000 and 3,000 nM had regions devoid of either apical-in or apical-out ZO1, indicating a loss of the tight junction barrier (Figure 4D). The optimal concentration range (30–300 nM) fits with the 190 ± 80 nM LPA concentration found in adult human CSF.¹⁵ To better understand the potential underlying mechanism behind apicobasal polarization, we performed a time series with samples fixed at 0, 2, 4, 8, 14, 18, and 48 h. Using our same metric of the percentage apical-out area, we determined that the organoids average 50% apical-out at approximately 5 h of 100 nM LPA exposure and greater than 90% by 18 h. The weakly ZO1-EGFP+ puncta seen at the 0-h time point appear to be apical primordial junctions known to contain ZO1 and be necessary for tight junction formation (Figure 4A).²¹ Over time, these puncta give way to either the apical-in lumens found in the BSA condition or apical-out homogeneous tight junction networks seen with LPA treatment (Figures 4C and 4E).

Figure 4. LPA dependency and reversibility of apical-out organization.

(A) Dose-response series for LPA for whole organoids.

(B) Higher magnification views from the panels in (A) highlight the intermediate effect at 3 nM, the most consistent apical-out structure from 30 to 300 nM, and obvious loss of ZO1 tight-junctions in some regions with 3,000 nM LPA treatment.

(C) Images of time series with vehicle (BSA) or LPA treatment at 0–48 h.

(D) Quantification of dose response data presented in (A and B) with four to eight organoids per group across two independent experiments. An agonist curve (black) was fitted to the data from 0 to 300 nM LPA. A quadratic equation was fitted to the data from 3 to 3,000 nM (blue) to show the inverted U. The mean LPA concentration and SD from human CSF are also shown as vertical dashed and dotted lines, respectively.

(E) Quantification of time series resented in (C) with three or four organoids per group from one experiment. An exponential plateau curve was fit to the data.

(F) At 96 h, the polarity persists based on the treatment; however, at 48 h, some organoids were switched from LPA-into BSA vehicle-containing media or from BSA into LPA with the resulting polarity (apical-in or apical-out) determined by the final treatment condition.

Organoids are day 7 of differentiation. Scale bars, 100 μm. Each data point is a single organoid.

To determine if the apical-out tight junctions formed are caused by increased ZO1 protein production, we performed immunoblot analysis for organoids treated with 24 h of LPA or S1P at 10, 100, and 1,000 nM concentrations. We saw slight increases in both 100 nM treatments (5% for LPA and 19% for S1P) with a greater than 30% decrease with 1,000 nM exposure for each (Figure S4). These data indicate that phospholipids are causing ZO1—and possibly other apical or tight junction proteins—to relocate to the outside surface of the organoids rather than increasing protein expression. This hypothesis is further corroborated by the reversibility of the phenotype. LPA treated organoids develop their polarity within 24 h and maintain their polarization to 96 h (Figure 4F). For some organoids, we swapped culture conditions after 48 h (BSA → LPA or LPA → BSA). In each case, we found the organoids had their stereotypical structure determined by their final media, indicating the LPA must be present for continued apical-out orientation and can re-orient apical-in organoids when ECM is absent (Figure 4F).

LPA and S1P cause apical-out polarity via GPCR/RhoA signaling

Given the specificity of LPA and S1P, we hypothesized that the apical-out polarity is likely due to GPCR signaling through several characterized LPA and S1P GPCRs. LPA has at least 6 GPCRs (LPAR1–6) and S1P has 5 (S1PR1–5), and all are involved in various canonical G-protein-activated pathways, including G12/13/RhoA/ROCK, GqPLC, Gi/Ras/MAPK, Gi/PI3K, and GS/cAMP. LPAR1 is highly expressed in the neural tube and developing cortex, and LPAR1 KO mice have exencephaly or craniofacial defects.²² Therefore, we tested the effect of the LPAR1/3 inhibitor KI16425. At an optimal LPA concentration (100 nM) to induce apical-out organoids, we found that 20 μM KI16425 completely blocks apical-out induction but had no effect on S1P-treated (100 nM) organoids as expected (Figures 5A–5F and 5S). From our previous RNA sequencing on day 6 organoids, we found high expression of LPAR1, LPAR2, and LPAR4 mRNA.¹⁷ Since this inhibitor only affects LPAR1 and LPAR3, LPA seems to affect polarity only through LPAR1 in our brain organoid system. These data also indicate that S1P does not function via LPAR1; however, we did not characterize the specific receptor involved since S1P is not expressed in CSF.^23,24 We further tested whether LPAR1 agonism alone could cause apical-out orientation using the LPAR1-specific agonist UCM-05194. We observed a partial effect of UCM-05194 at 0.2 μM, which is near the known median effective concentration (EC50 = 0.24 μM), and a more complete induction of apical-out organoids at 1 μM and 5 μM concentrations (Figures 5U and 5W).

Figure 5. LPA- and S1P-induced apical-out orientation via GPCR/RhoA.

(A–R) Neuroepithelial spheroids were incubated for 48 h with vehicle (BSA), 100 nM LPA or S1P and co-incubated with vehicle (A–C), LPAR inhibitor (KI16425, 20 μM) (D–F), Rho inhibitor (C3-transferase, 1 μg/mL) (G–I), Rho agonist (CN03, 1 μg/mL) (J–L), ROCK inhibitor (Y27632, 30 μM) (M–O), or SRF inhibitor (CCG-1423, 10 μM) (P–R).

(S) The percentage of each organoid that is apical-out was determined from the ZO1-EGFP images like those in (A–R) Comparisons between S1P and LPA only show a differential response with the LPAR inhibitor, KI16425, indicating its specificity for LPA receptors over S1P receptors. n = 3–5 organoids from one independent experiment with results replicated Figures S7 and S8. ****p < 0.0001 by two-way ANOVA with Sidak’s multiple comparison post-test.

(T) Apical-out quantification with various doses of the ROCK inhibitor, Y27632, show no response to 10 μM and significant blocking apical-out organoid formation with 30 and 100 μM concentrations. n = 4–11 organoids across two independent experiments. ****p < 0.0001 by one-way ANOVA with Brown-Forsythe and Welch multiple comparisons test.

(U) Images of organoids treated with listed concentrations of the LPAR1 selective agonist, UCM-05194.

(V) Organoids treated with LPA or cerebral spinal fluid and blocked with the Rho inhibitor, C3 transferase.

(W) Quantification of percent apical-out polarity for each organoid treated with LPAR1 selective agonist, UCM-05194. n = 2–4 organoids from one independent experiment. *p < 0.05, ***p < 0.001, ****p < 0.0001 by one-way ANOVA with Brown-Forsythe and Welch multiple comparisons test.

(X) Quantification of percent apical-out polarity for each organoid treated with Rho inhibitor, C3 transferase. n = 5–8 organoids from one independent experiment. ****p < 0.0001 by two-way ANOVA with Sidak’s multiple comparison post-test.

Organoids are day 7 of differentiation. Scale bars are 100 μm. Each data point is a single organoid. Data are represented as mean ± SD.

Based on these data and previous reports that showed G12/G13 expression were necessary for neural tube formation and closure,²⁵ we hypothesized that LPA and S1P induce apical-out orientation via the G_12/13/RhoA pathway downstream of their GPCRs. To test this hypothesis, we exposed organoids to LPA or S1P along with the Rho inhibitor C3 transferase. For either lipid at 100 nM, the Rho inhibitor, C3-transferase completely blocked the formation of apical-out organoids (Figures 5G–5I and 5S). The inhibitor was also effective when higher concentrations (1,000 nM) of LPA or S1P were used (Figure S5A). Treatment with the cell-permeable Rho activator, CN03, had no effect on LPA or S1P-dependent apical orientation (Figures 5K and 5L); however, the RhoA activator drastically altered the number, size, and shape of apical-in rosettes when LPA and S1P were not added (Figures 5J and S5B). In this condition, the number of rosettes was reduced (vehicle, 31 ± 5; CN03, 13 ± 1) and the size was dramatically increased (vehicle, 832; CN03, 1,970 median) appearing like the organoids treated with ECM (Figures 1J and S5C). These data indicate that Rho signaling from the outside of the organoid (LPA, S1P, or LPAR1 agonist) is necessary for apical-out polarization, while the Rho activator CN03 does not provide a polarized signal because it is cell permeable.

We next tested if phenotype was dependent on Rho effector Rho kinase (ROCK) by using the inhibitor, Y27632. It blocked the apical-out inducing effects of both 100 nM LPA and S1P at 30 μM but not at the typical 10 μM (Figures 5M–5O and 5T). This high level of ROCK inhibitor caused loss of both apical-out and apical-in structural formation. The images (Figures 5M–5O) look like the 0-h time point in Figure 4A. At this higher concentration, Y27632 is known to inhibit PRK2 and partially inhibit PKA and atypical PKC.²⁶ Therefore, our current data cannot prove that ROCK activity is necessary for the apical-out reorganization.

Actin polymerization is affected by several RhoA downstream effectors, including ROCK, PRK2, and mDia1.²⁷ To investigate if LPA affects F-actin polymerization, we generated apical-out organoids with 100 nM LPA and performed whole-mount phalloidin staining. We measured a consistent 2-fold increase in F-actin with 100 nM LPA exposure (Figure S6A). The addition of 10 μM Y27632 reduced F-actin to 41% of LPA alone but maintained apical-out polarity (Figure S6B). These data suggest that 100 μM LPA can increase filamentous actin possibly through ROCK but that this is not necessary for ZO1 tight junctions to form. Cofilin is an enzyme that severs F-actin but is inhibited by phosphorylation at serine-3 downstream of ROCK activity via LIMK1/2 leading to increased filamentous actin. Treatment with BMS3, a LIMK1/2 inhibitor, resulted in nearly complete cell death (data not shown); therefore, we assayed the activity of this pathway by measuring p(S3)-Cofilin after LPA or S1P exposure. Phosphorylation was only significantly increased after 1,000 nM of either phospholipid (Figures S6C and S6D). In a time-series using 1,000 nM LPA, we observed that this increase began after 1 h of exposure with no change in phosphorylation in the vehicle condition (Figure S6E). To directly test whether actin polymerization is important for apical-out polarization, we inhibited actin polymerization using cytochalasin B. The results were similar to the high concentration of ROCK inhibitor (30 μM) with many ZO1-EGFP puncta throughout which are likely primordial junctions, indicating a complete lack of polarization (Figure S6F). Therefore, it seems that some minimal level of actin polymerization is necessary for apicobasal polarization but that apical-out orientation is possibly dependent on a Y27632 off-target rather than ROCK.

Rho activation and F-actin polymerization cause serum response factor (SRF) to translocate to the nucleus and induce transcription of genes with serum response elements. To test whether this downstream transcription is necessary for LPA-induced apical-out orientation, we treated with the SRF inhibitor, CCG-1423 (10 μM). This inhibitor had no discernable effect on either the apical-in (BSA) or apical-out (LPA) orientation or morphology, indicating that SRF-dependent transcription does not play a role in apicobasal polarization (Figures 5P–5S). Replicate organoids for each of these condition in Figure 5 are presented in Figure S7 for the ZO1-EGFP-expressing line demonstrating robust phenotypes. Since this line was from a Japanese male donor, we tested the reproducibility of the LPA phenotype on organoids from a control iPSC line RPChiPS8023G1 that is from a Hispanic female donor. This line did not have the ZO1-EGFP fusion protein; therefore, we immunostained for ZO1. We found that this line also becomes apical out with 100 nM LPA and the effect is blocked by the Rho inhibitor (C3-transferase) and high concentration ROCK inhibitor (Y27632, 30 μM) (Figure S8).

We have alluded to a possible activation of this same pathway by LPA in CSF. Therefore, we obtained de-identified samples of CSF from patients with normal pressure hydrocephalus via the University of Michigan biorepository with institutional review board oversight and approval. Within 48 h exposure to CSF, all organoids became apical out, and this phenotype was blocked by the Rho inhibitor, C3-transferase (Figures 5V, 5X, and S9). Therefore, CSF activates this apical polarity pathway by the same Rho-dependent pathway and likely impacts organization of the developing ventricular zone and brain-CSF barrier.

LPA-dependent orientation maintained for more than 1 month with enhanced numbers and morphology of radial glia

To test whether the effects of LPA were transitory, we exposed brain organoids to 100 nM LPA continuously starting on day 5 and fixed samples each week for 5 weeks. Because these organoids were too large for whole-mount staining, we employed cryo-sectioning before immunostaining. At each time point, organoids without LPA had small apical-in lumens (Figures 6A–6D′) while those exposed to LPA remained apical out with ZO1-EGFP expressed contiguously on the outside edge. PAX6, a marker of forebrain radial glia, labeled most cells at each condition and time point. However, the LPA-treated organoids had more cells expressing PAX6, more consistent fluorescent intensity across PAX6-expressing cells, and a more radial orientation closely resembling a pseudostratified VZ-like morphology. LPA-treated organoids at day 35 had twice the number of PAX6⁺ cells compared with controls (LPA, 80 ± 20%; vehicle, 40 ± 10%). Notably, mice with LPA injected in the ventricles have been reported with an elevated number of TBR2⁺ intermediate progenitors in the developing cortex.²⁸ While TBR2⁺ cells peaked at day 21 in both organoid conditions, at day 35 LPA-treated organoids have three times as many TBR2+ cells (vehicle, 0.4 ± 0.2%; LPA, 1.4 ± 0.5%), likely due to the elevated number of remaining PAX6⁺ radial glia (Figure S10).

Figure 6. Long-term LPA causes persistent apical-out orientation and improved VZ morphology.

Organoids were treated with vehicle (A–D′) or 100 nM LPA (E–H′) starting on day 5. Organoids were cryosectioned and immunostained for PAX6 (red), TBR2 (white), ZO1-EGFP (green), and DNA (blue). Magnified views of insets in (A–H) are shown in (A′–H′). Scale bars, 100 μm.

We identified a large necrotic core in the center of each apical-out organoid at these later time points (Figures 6E–6H). From 48-h LPA organoid cryo-sections, we see a pseudostratified neuroepithelial-like layer that extends one-third to one-half of the radius of the organoids as shown by nestin or N-Cadherin staining (Figures S1B and S1C). While still expressing the markers of the anterior neural tube (FOXG1, PAX6, nestin, and N-cadherin), the center cells do not seem to connect to this layer or extend processes to the organoid surface. This core is strongly cleaved caspase-3 positive at 48 h (Figure S1D) likely due to isolation from media nutrients from the complete tight junction barrier surrounding the surface of the organoid (Figure S1A). While the radial glia can transport nutrients across their apical endfeet, delaminated cells cannot. We hypothesize that apical-out barrier formation could be the cause for the necrotic cores seen in brain organoid techniques that use KOSR-based media. These systems form temporary apical-out brain organoids (until KOSR removal) as shown by atypical PKC and N-cadherin in these publications.^11,12

LPA orients intestinal and lung organoids

Up to this point, our data indicate the role of LPA in apicobasal polarization in the neuroepithelium. We next asked whether LPA serves as a critical apical cue for other organoid systems. To investigate this possibility, we tested the effects of LPA on patient-derived lung and colon epithelial organoids, iPSC-derived multilineage intestinal organoids, and spheres of undifferentiated hiPSCs. Small airway epithelial organoids derived from adult patient donor lungs were expanded and removed from Matrigel according to Co et al. (2021).³ Lung organoids were resuspended in growth medium using DMEM/F12 with or without 100 nM LPA. We selected apical-in polarized lung organoids with an obvious single lumen into 96-well low-adherence dishes. After 5 days of 100 nM LPA exposure, motile cilia could easily be observed by light microscopy using phase contrast in all organoids along with organoid rotation. The number of organoids with cilia on the outer surface were counted. Only organoids exposed to LPA had any exterior cilia (23/48 with LPA, 0/49 in control media) (Figures 7A and 7B′ quantified in Figure 7G). It should be noted that apical-in lumens had motile cilia observable by phase microscopy that did not label with the acetyl-tubulin antibody (Figures 7A and 7B′). These organoids also expressed ZO1 and F-actin on their outer edge with LPA treatment only (Figures 7A and 7B′).

Figure 7. LPA causes apical-out orientation in non-neural organoid models.

(A and B′) Small airway organoids removed from ECM were placed in medium with vehicle or 100 nM LPA. Single confocal sections (A and B) and maximum projections (A′ and B′) from whole-mount organoids immunostained for apical proteins reveal apical-in orientation in nearly all vehicle-treated organoids and apical-out for nearly all LPA-treated organoids.

(C–F) Human intestinal organoids were placed in medium with vehicle or 100 nM LPA for either 7 (C and D) or 14 days (E and F). Staining for E-cadherin and CDX2 show the position of the intestinal epithelium.

(G) Quantification of motile cilia position in lung organoids by phase microscopy from one experiment. n = 49 and 48 organoids respectively from one independent experiment. ****p < 0.0001 by χ² test.

(H) Quantification of HIO orientation by E-cadherin-EGFP. n = 12 organoids for each treatment from one experiment. However, the LPA and S1P (and control data) were repeated in two to three additional experiments each with the same results.

(I) Proposed signaling mechanism for apical-out orientation for both brain organoids and intestinal organoids.

(J–K′) Undifferentiated hiPSC spheres were placed in media with vehicle or 100 nM LPA. Single confocal sections (J and K) and maximum projections (J′ and K′) from whole-mount spheres labeled by ZO1-EGFP demonstrated either apical-in for vehicle-treated spheres or apical-out for LPA-treated spheres. Scale bars, 100 μm.

Patient-derived colon organoids (called colonoids) were also generated and removed from Matrigel before being suspended in media alone or with 100 nM LPA. We found colonoids in media were mostly mixed polarity with ZO1 and F-actin on the outside surface and on internal lumens. This results from having a bilayer of cells that yield dysmorphic, bumpy organoid shapes (Figures S11A–S11A′). LPA-treated colonoids had spherical organoids that were a single cell layer thick (Figure S11B). ZO1 and F-actin label the outside of the organoids with E-cadherin-labeled adherens junctions perpendicular to the outside apical surface. The LPA-treated organoids also had stronger ZO1 fluorescent signal and had bulging cells indicative of goblet cells at a time point that typically does not have this level of maturity (Figure S11B′). Unfortunately, the phospholipid-free colonoids often died or fused together, making robust, reproducible quantification difficult. Therefore, we switched to using multilineage human intestinal organoids from iPSCs previously developed by our group.²⁹ These organoids have epithelial, mesenchymal, and even endothelial cell types with a vascularized structure. We skipped the embedding step into Matrigel domes and instead tested the effects of LPA and S1P on suspended organoids. We then immunostained sections for the epithelial markers E-cadherin and CDX2. Similar to previous publications, the suspended organoids lacking LPA or S1P had apical-in epithelial structures encased in mesenchymal and endothelial cells. Treatment with LPA or S1P at 10 μg/mL (23 and 26 μM, respectively) caused inversion in every organoid we imaged: apical-out epithelium on the outer edge of the organoid encasing the other cell types (Figures 7C–7F and S12). This higher concentration of phospholipids needed for the inversion may be necessary to penetrate the cells surrounding the epithelial layer and cause migration of these cells to the outer edge. We tested whether the same GPCR/RhoA pathway was involved by co-incubating with the inhibitors, KI16425 and C3-transferase. While Rho inhibitor blocked flipping more than 80% of organoids from both LPA and S1P, the LPAR inhibitor only blocked flipping with the LPA exposure (LPA + KI, 1/12 inverted; S1P + KI, 11/12 inverted) (Figure 7H). Example whole-mount images for each condition are shown in Figure S12. Thus, the phospholipids cause the inverted polarity via the same mechanistic pathway in the neural tube (Figure 7I).

Because human pluripotent stem cells form epithelial-like colonies in their primed state, we tested whether LPA could cause apical-out orientation in hiPSC spheres. Using the same paradigm as the neuroepithelial cell spheres, we placed Aggrewell-generated hiPSC spheres in mTeSR-Plus media alone or with 100 nM LPA added. All organoids treated with LPA formed ZO1-EGFP-labeled apical-out organoids while mTeSR-plus alone resulted in apical-in lumens (Figures 7J–7K′). These data demonstrate that LPA is a broad apical cue in the absence of ECM for multiple cell lineages and developmental time points, and that LPA/S1P exposure are likely the cause of many—if not all—apical-out organoid systems.

DISCUSSION

In this study, we show that apical-out orientation in ECM-free brain organoid culture is induced by two phospholipids found in serum: LPA and S1P. These phospholipids act as exogenous apical cues that can be used to orient apicobasal polarity in the absence of exogenous ECM and function by activating their respective GPCRs. We also find that this pathway involves the downstream activation of Rho, which is necessary for apical-out brain organoid orientation. Notably, the different culture media that we have identified in the literature as producing apical-out organoids, including for intestinal and lung organoids, contain LPA.^{3–6,13,30,31} Given that both phospholipids are present in amniotic fluid but only LPA is found in CSF at concentrations that induced apical-out organoids in our assay, we propose that both LPA and S1P are critical for the induction of neuroepithelial polarity in the neural plate/tube, and that LPA in CSF is necessary for maintaining polarity and barrier function in the ventricles.

Previous work has shown that apical-out inversion in MDCK cells induced by incubation with beta1-integrin-blocking antibody is RhoA dependent.³⁰ This inversion could also be produced by hanging-droplet suspension.³¹ These ECM-free cultures used fetal bovine serum (FBS) at 10% in the media. FBS has an average LPA concentration of approximately 5 μM resulting in approximately 500-LPA in the growth media.³² Therefore, LPA from serum likely resulted in the polarity inversion seen in these cultures. Furthermore, apical-out polarity inversion of suspension micropapillary carcinoma cells is known to be Rho/ROCK dependent.³³ In each case, the cells were grown in suspension in media containing serum.^30,31,33 The use of LPA- or S1P-containing media under basal conditions for most cell culture systems (either in serum, KOSR, or Albumax containing medias like advanced DMEM/F12) is likely the reason that LPA and S1P have not been previously identified as apical cues. Therefore, our study connects the known RhoA pathway activation necessary for apical-out orientation to LPA and S1P found in these medias. Currently, our data do not conclusively implicate ROCK in brain organoid polarization as these previous studies had shown. Perhaps this is due to our brain organoid model cells lacking polarization rather than inverting polarity. ROCK inhibition can block cytoskeletal remodeling necessary for inversion but perhaps not be essential for apical junction polarization and formation. One prior model of apical junction formation in bronchial epithelial cells found that PRK2 was the RhoA downstream effector necessary, not ROCK.³⁴ While we have tested a PRK2 inhibitor, this compound was lethal to our cultures at concentrations typically used (data not shown). Future experiments using knockdown rather than pharmacological inhibition may be needed to further elucidate the signaling events downstream of RhoA necessary for apical polarization in our model system.

In the context of neural tube formation and closure, apicobasal polarity is necessary for proper formation, since the rounded tube forms due to polarized constriction of the apical surface of each cell. KO of each part of the pathway, including the LPA-producing enzyme (autotaxin/ENPP2), LPA receptor (LPAR1), G13 (GNA13), and RhoA lead to loss of neural tube closure and/or exencephaly.^22,35–38 LPA is expressed between 100 and 200 nM in CSF (and serum) where it is produced in the choroid plexus by autotaxin.¹⁵ This LPA concentration was optimal in our brain organoid system to induce complete apical-out polarity. In contrast, S1P is exceptionally low in CSF (1–2 nM) and high in serum (400–1,000 nM).^24,39 Loss of S1P production in mice from SphK1 and SphK2 KO also resulted in lack of neural tube closure despite low S1P in CSF.⁴⁰ The concentration of LPA and S1P in amniotic fluid are approximately 27 and 20 nM, respectively,^41,42 and based on our apical-out induction with 30 nM LPA, this should be sufficient to provide apical signaling. Therefore, LPA and S1P in amniotic fluid may both be needed for proper neural tube closure, but only LPA is needed in CSF for proper barrier maintenance at the CSF/VZ interface. In the future, amniotic fluid should be tested for its capacity to induce apical-out organoid formation.

When Enpp2, SphK1/SphK2, or GNA13 are knocked out, vasculature fails to form, highlighting the importance of this signaling pathway for tube formation beyond the neural tube alone.^35,36,40 Since these genes are all involved in LPA/S1P synthesis or signaling, these studies suggest that LPA and S1P are important for neural tube and vasculature formation, even in the presence of ECM proteins. The importance of these phospholipids seems highly conserved. For example, loss of the sphingolipid producing enzyme in C. elegans results in the formation of many circular intestinal lumens rather than one contiguous tube.⁴³ Therefore, although ECM-integrin signaling is sufficient for apicobasal polarization, loss of a phospholipid apical instructive cue may lead to loss of tubulogenesis due to lack of contiguous apical membranes.

Our optimal concentration for apical-out orientation of 100 nM LPA is lower than the typical concentrations used to elicit LPA-dependent phenotypes, which are typically 1–10 μM.^14,44 The approximately 0.5 μM LPA in typical 10% serum containing media.³² Interestingly, plasma has a concentration range of 0.7–80 nM, and the elevated LPA in serum is due to platelet activation.⁴⁵ We observe disruption of the tight junction barrier formation and excessive F-actin in the high concentration range (1–10 μM). These disruptive effects of elevated LPA in our model may explain why overexpression of autotaxin leads to loss of neural tube closure and injection of excess LPA in ventricles, leading to the loss of ventricular barrier integrity.^46,47 Therefore, animal models are congruent with our dose-response data that too much or too little LPA is detrimental to the formation and stability of the neural tube.

The results of this study also have implications for cancer biology where polarity switching often occurs in metastatic cancers in the blood stream. One report describes blockade of cancer cell polarity switching by Rho/ROCK inhibitors applied in suspension culture.³³ Given our findings, this polarity switching may be induced by apical cue phospholipids, LPA and S1P, present in blood or cell culture media. This knowledge may provide additional targets for treating metastatic cancers.

Reproducible production of apical-out brain organoids opens many future research possibilities. As we have shown, the apical surfaces can be imaged in whole-mount organoids at high magnification allowing for better understanding of protein localization at the tight junctions and apical membrane in neural tube organoids. These organoids will allow for investigation of primary cilia, growth factor receptors, and nutrient absorption that occur from the CSF to the apical membranes of neuroepithelial cells, followed by those of radial glial and ependymal cells that line the brain ventricles at different stages of development. These organoids are also useful models for studying CSF infections by providing direct access to the interfacing surface.

Apical-out intestinal and lung organoids are crucial for studying nutrient or oxygen absorption, microbiome, and pathogen interactions.^7,48 Although prior published apical-out organoid techniques in these tissues used LPA signaling found in their medias,^3,5 our results will likely increase the efficiency of these techniques. We also present a unique apical-out human intestinal organoid system offering new opportunities to study apical-specific intestinal functions in a multilineage context. Furthermore, our apical-out organoid system avoids the use of expensive ECM hydrogels and difficult removal processes reducing cost and labor.

Limitations of the study

We infer that serum and CSF caused apical-out orientation via LPA signaling. The known standard concentrations of LPA in these fluids are sufficient to cause apical-out orientation; however, we did not measure the concentrations of LPA or S1P within these fluid samples. Furthermore, we used normal pressure hydrocephalus patient CSF since these patients produce large amount of CSF that must be removed via shunt. CSF is difficult to obtain from normal healthy controls. We assume similar phospholipid concentrations in these patients compared with reported concentrations for normal controls. We also did not investigate all potential phospholipid species that may be present in these fluids. In our reversibility data, we did not test whether the apical-out orientation remains stable if LPA was removed after more than 96 h. It is possible that the orientation becomes stable independently of ongoing phospholipid presence.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Andrew Tidball (atidball@umich.edu).

Materials availability

The SHROOM3-KO isogenic control iPSC line and the intestinal and lung patient organoid lines can be requested from the lead contact. No other novel plasmids, constructs, or reagents were used in this study.

Data and code availability

• Microscopy and western blot data reported in this paper will be shared by the lead contact upon request.

• Original code for apical-out percent image analysis has been deposited in GitHub. A publicly available link can be found in the key resources table.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-Acetyl-tubulin (1:1000)	Sigma	Cat#T6793; RRID: AB_477585
Mouse monoclonal anti-Arl13B (1:200)	NeuroMab	Cat#N295B/66; RRID: AB_2877361
Mouse monoclonal anti-β-Catenin (1:300)	BD Biosciences	Cat#610153; RRID: AB_397554
Rabbit polyclonal anti-Cingulin (1:1000)	Invitrogen	Cat#PA5–55661; RRID: AB_2639733
Rabbit monoclonal anti-Cleaved Caspase-3 (1:1000)	BD Biosciences	Cat#559565; RRID: AB_397274
Mouse monoclonal anti-CDX2 (1:500)	BioGenex	Cat#MU392A-UC; RRID: AB_3101998
Goat polyclonal anti-E-Cadherin (1:500)	R&D Systems	Cat#AF748; RRID: AB_355568
Mouse monoclonal anti-E-Cadherin (1:1000)	BD Biosciences	Cat#610182; RRID: AB_397581
Mouse monoclonal anti-GAPDH (1:1000)	Sigma	Cat#G8795; RRID: AB_1078991
Rabbit monoclonal anti-MyoIIB (1:1000)	Cell Signaling	Cat#8824; RRID: AB_11217639
Rabbit polyclonal anti-Myosin IIB (1:1000)	Biolegend	Cat#909901; RRID: AB_2565101
Mouse monoclonal anti-N-Cadherin (1:200)	Thermo Fisher	Cat#33–3900; RRID: AB_2313779
Mouse monoclonal anti-Nestin (1:300)	Millipore	Cat#MAB5326; RRID: AB_2251134
Rabbit polyclonal anti-MPP5/PALS1 (1:100)	Proteintech	Cat#17710–1-AP; RRID: AB_2282012
Rabbit polyclonal anti-Partitioning-defective 3 (PAR3) (1:250)	Millipore	Cat#07–330; RRID: AB_2101325
Rabbit polyclonal anti-PAX6 (1:1000)	MBL	Cat#PD022; RRID: AB_1520876
Rabbit monoclonal anti-Phospho-Cofilin-(S3) (1:1000)	Cell Signaling	Cat#3313; RRID: AB_2080597
Mouse monoclonal anti-Phospho-Vimentin (S55) (1:1000)	MBL	Cat#D076–3; RRID: AB_592963
Mouse monoclonal anti TBR2(EOMES) (1:400)	R&D	Cat#MAB6166; RRID: AB_10919889
Rabbit polyclonal anti-TPX2 (1:500)	Novus	Cat#NB500–179; RRID: AB_10002747
Rabbit polyclonal anti-ZO1 (1:1000)	Thermo Fisher	Cat#61–7300; RRID: AB_2533938
Mouse monoclonal anti-ZO1 (1:200)	Thermo Fisher	Cat#33–9100; RRID: AB_2533147
Chemicals, peptides, and recombinant proteins
ACA	Medchem Express	Cat#HY-118628
C3-Transferase	Cytoskeleton, Inc.	Cat#CT04
CCG-1423	Medchem Express	Cat#HY-13991
Cytochalasin B	Cayman Chemical	Cat#11328
KI16425	Selleck Chem	Cat#S1315
Rho Activator II	Cytoskeleton, Inc.	Cat#CN03
UCM-05194	Cayman Chemical	Cat#41682
Y27632	Chemdea	Cat#CD0141
Lysophosphatidic acid	Sigma	Cat#L7260
Cardiolipin	Sigma	Cat#C0563
Lysophosphatidylcholine	Sigma	Cat#L4129
Lysophosphatidylethanolamine	Sigma	Cat#L4754
Phosphatidic Acid	Sigma	Cat#P9511
Phosphatidylcholine	Sigma	Cat#P3556
Phosphatidylethanolamine	Sigma	Cat#P7943
Phosphatidylinositol	Sigma	Cat#P0639
Phosphatidyl-L-serine	Sigma	Cat#P7769
Sphingosine-1-phosphate	Cayman Chemical	Cat#22498
Experimental models: Cell lines
WTC-mEGFP-TJP1-cl20 (mono-allelic tag)	Coriell Institute Biorepository	ID#AICS-0023 RRID: CVCL_JM18
WTC-mEGFP-CDH1-cl32 (bi-allelic tag)	Coriell Institute	ID#AICS-0114–032 RRID: CVCL_C1XK
Human iPSC 802–3G	ReproCell (via Synthego)	CAT#RPChiPS8023G1 RRID: CVCL_E3R4
SHROOM3-KO isogenic WT iPSC line	Takla et al.17	N/A
iPSC72.3	Cellosaurus49	Cat#CCHMCi001-A; RRID: CVCL_A1BW
WAe001-A hESC	WiCell	Cat#wa01-cgmp-material RRID: CVCL_9771
Colon-88 patient-derived colon organoid line	Dame et al.50	N/A
HT617 patient-derived airway organoid line	Gift of Life, 20 yo female	N/A
Software and algorithms
CellPose	Stringer et al.51	cellpose.com
Prism 10	GraphPad Software	Graphpad.com
FIJI	Open Source	FIJI.sc
Image Studio Lite Software	LI-COR BioTech	Licorbio.com
Brain Organoid Apical-out % pipeline	https://github.com/MaggieCoder/Neuroepithelial-Organoid-Analysis-Pipeline	N/A

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Human induced pluripotent stem cell lines

We purchased the AICS-0023 cell line from the Coriell Institute Biorepository (Camden, NJ, USA) in the Allen Cell collection. This line contains a monoallelic mEGFP-TJP1 (which encodes for the zona occludens-1 [ZO1] protein) to label the lumen in live SOSR-COs. This line is from a male donor of Japanese descent. We have previously published a control iPSC line generated from commercially available foreskin fibroblasts that we used for experiments utilizing Phalloidin-Alexa488 to measure f-actin.¹⁷ Key phospholipid and inhibitor data was confirmed in the commercially available control iPSC line RPChiPS8023G1 from ReproCell, from a female donor of Hispanic descent. Human intestinal organoids (HIOs) were differentiated from the AICS-0114 cell line from the Coriell Institute Biorepository (Camden, NJ, USA) in the Allen Cell collection. This line contains a biallelic mEGFP-CDH1 (which encodes for E-Cadherin protein) from the same male Japanese donor as AICS-0023. All lines were tested for mycoplasma contamination. These lines were used with regulatory approval from the University of Michigan Human Pluripotent Stem Cell Research Oversight (HPSCRO) Committee. Since these lines are deidentified, this study is exempt from Human Subjects research and IRB approval. The EGFP expression pattern was sufficient to authenticate the identity of the tagged lines.

Human cerebrospinal fluid samples

Deidentified aliquots of cerebrospinal fluid from 3 normal pressure hydrocephalus patients were obtained from the University of Michigan biorepository. IRB-MED determined these samples were exempt from human subjects research. The CSF from these 3 patients were mixed at equal volumes for form a pooled sample. The entire 100 μL media volume used for incubating brain organoids was replaced with CSF.

METHOD DETAILS

iPSC culture

The iPSC cultures were maintained on 6-well TC dishes coated with Geltrex (Thermo, Waltham, Ma, USA) diluted 1:200 dilution in DMEM/F12 medium (Thermo) at 37°C. Cells were cultured in mTeSR1 or mTeSR-Plus medium (Stemcell Technologies, Vancouver, BC, Canada). When the colonies reached ~40% confluency, the cultures were incubated with 1 mL of hypertonic solution containing sodium citrate and potassium chloride for 2 min. The solution was then replaced with culture medium, and the colonies were mechanically detached with a mini cell scraper. The solution was then pipetted up and down 3–6 times to break the colonies into smaller pieces. The colony fragment suspension was then re-plated at a dilution of 1:8 onto newly Geltrex-coated plates.

ECM-free organoid differentiation

We used Accutase to passage 1 × 106 iPSCs onto Geltrex-coated (1:50 dilution in DMEM/F12) 12-well plates in mTeSR with 10 μM Y-27632 (Tocris, Bristol, UK). When the cells reached 80–100% confluency, the medium was changed to 3N (50:50 DMEM/F12:neurobasal with N2 and B27 supplements [Thermo]) without vitamin A with 2 μM DMH-1 (Tocris), 2 μM XAV939 (Cayman Chemical, Ann Arbor, MI, USA), and 10 μM SB431542 (Cayman Chemical) with 2 mL of medium per well. Subsequently, 1.5 mL medium were changed daily with 1 μM cyclopamine (Cayman Chemical) beginning on day 1 along with the other 3 inhibitors. On day 4, the monolayer was dissociated by incubating with Accutase for 8 min. After centrifugation, the cells were resuspended in media with the 4 inhibitors and 10 μM Y-27632. 2 × 106 cells were added to each well of a 24 well Aggrewell 800 in 2 mL of media, and the plate was centrifuged at 100g for 3 min. The following day, individual spheres were transferred to wells of a low-adherence U-bottom 96 well plate in 3N-A media with 3 μM CHIR99021 and test compounds such as LPA.

Immunocytochemistry

Organoids were fixed in 4% paraformaldehyde in PBS for 1 h at 4°C. For cryosectioning, the organoids were embedded in OCT, and 15 μm sections were attached to slides. After PBS washes to remove OCT, sections were permeabilized with 0.1% Triton X-100 for 20 min. Each slide incubated with ICC buffer with phosphate-buffered saline, with 5% normal goat serum, 1% BSA, and 0.05% Tween 20. Primary antibodies were diluted in ICC buffer according to the dilutions in key resources table and incubated overnight at 4 C. Each was washed 4x in PBS +0.05% Tween 20 for X min each. Secondary antibodies (either Goat anti-rabbit-Alexa568 or Goat anti-mouse-Alexa647) were diluted 1:1000 in ICC buffer and incubated for 90 min. The sections were then counterstained with Hoescht DNA dye for 10 min and coverslips mounted using Glycergel. For wholemount, organoids were permeabilized and blocked in ICC buffer with 0.1% Triton X-100 over night at 4°C on a Nutating shaker. Primary antibodies were added in this same ICC buffer with TX-100 overnight at 4°C on a Nutating shaker. The organoids were washed 4 × 15 min in PBS with 0.05% Tween-20. Secondary antibodies were added at 1:1000 dilution in ICC buffer with TX-100 at 4° C on a Nutating shaker. After 4 additional washes in PBS with Tween-20, the organoids were placed thin bottom 96 well plates in PBS for imaging. Microscopy was performed with an Andor BC43 confocal microscope with 20x objective. Z-series of the bottom 100 μm for each organoid were obtained with a 1.5 μm step size (67 images each). Maximum projections were obtained with the same brightness and background subtraction applied for each experiment.

Immunoblotting

Total protein was extracted using CelLytic^™ M Cell Lysis Reagent (Sigma-Aldrich, #C2978) with PhosSTOP^™ (Roche, #04906845001) and cOmplete^™ ULTRA Protease Inhibitor Cocktail (Roche, #05892970001). Protein concentrations were determined by Qubit^™ Protein BR Assay kit (Invitrogen, #A50668). Equal amounts of protein were denatured at 99° C for 5 min, separated by SDS-PAGE, and transferred onto NC membranes. The blots were blocked with 5% nonfat dry milk in Tris-buffered saline with 0.1% Tween 20 (TBS-T) for 1 h, followed by incubation with primary antibodies at 4°C overnight (Antibodies and dilutions in key resources table). The blots were washed with TBS-T followed by incubation with IRDye 800 CW goat anti-rabbit or goat anti-mouse secondary antibodies (LI-COR Biosciences, 1:8000) for 1 h. After additional washes with TBS-T, the immune complexes were detected by the Odyssey Imaging Systems (LI-COR Biosciences). Image Studio TM Lite Software (LI-COR Biosciences) was used to quantify the protein signals.

Lipid solutions

Lyophilized lipid powders were dissolved in soluble organic solvents. The solutions were aliquoted into glass culture test tubes to achieve 0.3 mg phospholipid each. The solvent was then removed using a Savant Speed Vac with solvent collection flask for 0.5–3 h until only a phospholipid film or small particulates remained on the glass. The film was then dissolved into a 10% fatty acid free BSA solution (catalog number) with warming at 37 C for 30 min followed by vortexing and additional warming as needed until fully dissolves to a concentration of 0.3 mM. The solutions were then sterile filtered through a low protein binding 0.45 μm syringe filter. LPA was also dissolved in DMSO and found to have no difference in orientation compared with the BSA solution (Figure S3).

Patient-derived lung organoid culture and ECM removal

HT-617 airway organoids were maintained in Matrigel with FANY media. Organoids were move to a 15 mL conical tube with 10 mL EDTA and rotated in cold room for 1 h. The organoids were pelleted at 300 g for 3 min. The EDTA was aspirated and the pellet washed with 5 mL of DMEM/12. The organoids were pelleted at 300 g for 3 min. The DMEM/F12 was removed and the organoids were resuspended in FANY media with the BSA replaced with fatty acid free BSA. Organoids with a clear apical lumen were picked into a low-adherence U-bottom 96 well plate the FAFBSA containing FANY with or without 100 nM LPA. The cultures were fixed and whole mount immunostained.

Patient-derived colonoid culture and ECM removal

Colon-88 colon organoids were maintained in Matrigel with L-WRN Conditioned Media for 6 days prior to harvesting. The organoids were transferred to a 50 mL conical with 45 mL of 5 mM EDTA and rocked in a cold room for 1 h. Organoids were pelleted at 300 g for 3 min. The EDTA was removed, and the pellet was washed in 15mL DMEM/F12. Organoids were pelleted at 300 g for 3 min. Organoids were resuspended in 0.567 mL of DMEM/F12. 81 μL of resuspended organoid mixture was transferred to 6 wells of a low-adherence 6 well plate. The wells contained Advanced DMEM/F12, DMEM/F12, and DMEM/F12 + 100nM LPA, with 2 wells per condition. After 24 h, one of each well was fixed and whole mount immunostained. After 48 h, the remaining wells were fixed and whole mount immunostained.

Human intestinal organoid culture and slide preparation

Human Intestinal Organoids (HIOs) were generated from iPSCs (line iPSC72.349) by aggregation as previously described,^52,53 with the exception that a 96-Well U Bottom plate was used. Media was changed every 3–4 days. HIOs were placed in 10% Neutral Buffered Formalin at room temperature on a rocker for 24 h to allow for complete fixation. After fixation, HIOs were washed three times in UltraPure DNase/RNase-Free Distilled Water for 1hour each. The tissue was dehydrated through a methanol series diluted in UltraPure DNase/RNase-Free Distilled Water for 60 min per solution in the following order: 25% MeOH, 50% MeOH, 75% MeOH, 100% MeOH. Dehydrated tissue was placed in 100% EtOH, followed by 70% EtOH, and then perfused with paraffin using an automated tissue processor (Leica ASP300) with solution changes every hour overnight. Tissue was subsequently placed into tissue cassettes and base molds for sectioning. Sections, 5 μm-thick, were cut from paraffin blocks onto charged glass slides. Slides were dried for 1 h at 60°C and used within a week for immunofluorescence staining. Slides were deparaffinized with two 5-min washes in Histo-Clear II, followed by rehydration through a graded ethanol series: two 2-min washes each in 100% EtOH, 95% EtOH, 70% EtOH, 30% EtOH, and finally two 5-min washes in ddH2O. Antigen retrieval was carried out in 100 mM trisodium citrate with 0.5% Tween 20 pH 6.0. Slides were placed in a pressure cooker for ~30 min, cooled, and washed three times in ddH2O for 5 min each. Slides were subsequently immunostained as described previously.

QUANTIFICATION AND STATISTICAL ANALYSIS

Apical-out quantification

ZO1-EGFP expressing brain organoid maximum projection images were manually quantified by a blinded, trained observer. In ImageJ, segmented polygons were used to measure the area of the total organoid and apical-out area. Only regions with >1 fully enclosed apical cell surface were considered. To confirm our results, we developed a fully automated apical-out segmentation tool in Python. This pipeline identifies the total organoid area and apical-out area by using an Otsu intensity threshold method and cell size threshold. When comparing this method to our manual quantification on the LPA concentration curve, we obtained a R² 0.986 and a slope of 0.942 indicating high concordance between manual and automated quantification.

Apical-in lumen metric quantification

For comparing apical-in lumens, a FIJI macro was used to generate a binary mask, hole fill, and particle measurement for particles >100 μm² and >0.2 circularity. The number of lumens, area, and roundness measurements were exported.

Cell type quantification

For quantifying the number of PAX6 and TBR2 expressing cells from the long-term LPA treated brain organoid cryosections, we utilized Cellpose with a user determined average nuclei diameter of 20 μm. This program was used for bis-benzamide, PAX6, and TBR2 labels. The PAX6 and TBR2 cell numbers for each image were divided by the bis-benzamide nuclei number to obtain percentages.

Statistical analysis

All graphs and statistical analysis were performed in PRISM 10 (GraphPad). We used non-parametric Kruskal-Wallis for comparing 2 groups and Brown-Forsythe and Welch ANOVA with Dunnet’s multiple comparison test for groups >2. Multivariate data was analyzed by two-way ANOVA with Sidak’s multiple comparison post-test. Count data analyzed by Chi-square test. All datapoints are from individual organoids with means and standard deviations depicted. Significance for any of these tests is defined as p < 0.05. Throughout the figures, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. For all brain organoid, iPSC spheroid, and human intestinal organoid (HIO) data, the results were nearly the same in all replicate organoids (3–6/condition) across 2+ experiments. Given the nearly binary apical-out phenotypes for many conditions, quantifications are often presented from a single independent experiment. However, these results were reproduced in additional experiments for each.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.115842.

Highlights.

• LPA/S1P induce apical-out brain, lung, and intestinal organoids

• LPA/S1P-dependent polarization occurs via GPCR/RhoA signaling

• Phospholipids in serum and CSF induce apical-out polarity

• LPA/S1P in media contribute to apical-out orientation in existing organoid systems