WNT Mimetic-Induced Lacrimal Gland Regeneration Reverses Aqueous Tear Deficiency

Abstract

Purpose: Dry eye disease is characterized by abnormal tear film composition and inflammation. Disruptions in the presence or secretory function of acinar cells can result in dry eye, leading to discomfort, damage, and vision loss. Although tear replacement and anti-inflammatory treatments have been investigated thoroughly, a method to induce epithelial restoration is lacking. Here, we asked whether WNT signaling activation via an antibody-based WNT mimetic platform might activate lacrimal gland acinar cells and restore tear secretion.

Methods: Primary murine lacrimal gland cells were used to establish three-dimensional acinar cell organoids. Dry eye disease was modeled in mice using a lacrimal gland excretory duct ligation. Transcriptional and cellular changes were investigated using single-cell sequencing.

Results: Frizzled and LRP5/6 receptors, which are essential for WNT signal transduction, are expressed in glandular acinar cells in vivo. We showed that murine acinar cells can be expanded as organoids by using WNT signaling activation. Here, we demonstrate that WNT signaling is an essential factor for acinar cell proliferation ex vivo. In a mouse model of dry eye disease, and intra-lacrimal gland treatment with a WNT mimetic targeting FZD1, 2, and 7 reversed aqueous tear deficiency. After excretory duct ligation damage, WNT mimetic treatment promoted acinar cell restoration and increased detectable tear volume production.

Conclusions: We demonstrate a role for WNT signaling in acinar cell proliferation. Our findings extend the potential for WNT pathway activation via a ligand mimetic platform to lacrimal gland regeneration. Pathway activation results in adult acinar cell proliferation in vitro and in vivo and tissue recovery.

Translational Relevance: Our novel WNT mimetic platform offers a promising alternative to symptom-focused treatments by actively stimulating acinar cell proliferation and restoration.

Introduction

Lacrimal glands, exocrine glands that secrete the aqueous component of the tear film, play essential roles in eye health and function. These glands consist primarily of specialized epithelial cells; ductal cells, secretory acinar cells, and myoepithelial cells.¹ Dysfunction of these cells caused by damage or inflammation can result in insufficient tear fluid production, as well as ocular pathologies.² People with dry eye disease experience irritation, pain, and potential damage to the ocular surface.³ Although artificial tears and anti-inflammatory eyedrops may relieve symptoms, they cannot restore glandular function.^4,5 Therefore, approaches to stimulate acinar cell recovery and lacrimal gland tissue regeneration remain a topic of great interest.⁶ 

The WNT/β-catenin signaling pathway plays key roles in the development and tissue homeostasis of mammalian organs.⁷ In glandular tissues, WNT proteins are involved in self-renewal, cell fate determination, and branching morphogenesis.^8,9 WNT signaling components have also been described in developing and adult lacrimal glands.^10–13 Enhanced pathway activation by R-spondin (RSPO) has been used in the expansion of ductal cell populations as organoids from both mouse and human tissue.¹⁰ The expansion of adult acinar cells remains challenging. Other lacrimal gland cultures are short lived,^14,15 expanded in serum containing medium,¹⁶ or generated from undifferentiated cells.^17,18 To our knowledge, no chemically defined culture protocol is available currently to expand lacrimal gland acinar cells. The effects of WNT on acinar cell turnover and functional tear production in vivo remain poorly understood, thereby hampering the development of therapeutic options. Recent efforts have identified novel WNT-activating approaches with characteristics of natural ligands,¹⁹ as reviewed in a previous article.²⁰ These synthetic WNT signaling molecules potently activate the pathway and can be applied readily in vivo.²¹ WNT mimetic technology has been demonstrated to stimulate tissue regeneration in several adult organs and has entered clinical studies recently.²⁰ 

Here, we explored the ability of our antibody-based WNT mimetic platform²¹ to stimulate lacrimal gland tissue regeneration. Pathway activation through FZD_1,2,7 enabled murine acinar cell expansion as organoids in vitro and accelerated tear secretion restoration in a model of dry eye in vivo. Through single-cell messenger RNA (mRNA) sequencing of treated glands in a mouse model of dry eye, we uncovered cell populations unique to the damaged environment and responsive to WNT mimetic treatment. The WNT-activating molecule described herein may serve as a basis for future therapeutic development. 

Methods

Protein Production

All recombinant proteins were produced in Expi293F cells (Thermo Fisher Scientific, Waltham, MA, USA) by transient transfection. The Fv-Fab proteins were first purified using cOmplete His-tag purification resin (Sigma-Aldrich, St Louis, MO, USA); other proteins were first purified using MiniChrom Mab-Select SuRe (Repligen, Waltham, MA, USA), unless otherwise specified. All proteins were further polished with Superdex 200 Increase 10/300 GL (GE Healthcare Life Sciences, Marlborough, MA, USA) size-exclusion chromatography using 1× HEPES-buffered saline buffer (20 mM HEPES pH 7.4, 150 mM NaCl). The proteins were subsequently examined by SDS-polyacrylamide electrophoresis and estimated to have greater than 90% purity. 

Murine Lacrimal Gland Cultures

All adult stem cell–derived organoids were maintained at 37°C in a 5% CO₂ environment. Freshly isolated adult female murine lacrimal glands were used for adult stem cell–derived organoid outgrowth. Lacrimal glands were excised and chopped into smaller pieces using a scalpel and forceps. Tissue fragments were digested using collagenase, hyaluronidase and DNase in DMEM at 37°C shaking for 20 minutes in a total of 5 mL. Digested fragments were dissociated mechanically further by pipetting up and down. The near single-cell suspension was washed and pelleted. On average one dissociated gland was plated in five 20-µL Matrigel droplets. If more material was needed, multiple glands were pooled in the same procedure. Matrigel droplets were solidified for 15 minutes prior to addition of expansion medium consisting of Advanced DMEM, Glutamax, Antibiotic-antimycotic, HEPES, B27 supplement, N2 supplement, epidermal growth factor (EGF), fibroblast growth factor 2 (FGF2), and FGF10. 

Lacrimal Gland Single-Cell Dissociation

Adult murine lacrimal glands were excised and stored in 1 mL cold DMEM until further processing. Individual glands were placed in a dry Petri dish and chopped into small pieces (approximately 1 mm³) using a scalpel and forceps. The pieces of tissue were transferred into 5 mL of digestion buffer containing collagenase, hyaluronidase, and DNase1 in Advanced DMEM/F12. After 20 minutes shaking at 37°C, the suspension was pipetted up and down a few times before passing through a 100-µm filter. The cell suspension was washed once in Advanced DMEM/F12 and further digested in 1 mL of TrypLE for 5 minutes. Further mechanical shearing using a small pipette tip could further reduce the larger fragments of cells. The final cell suspension was passed through a 30 µm filter prior to further processing on a fluorescent activation cell sorter or by magnetic bead isolation. 

Immunofluorescence and In Situ Hybridization

For organoids staining, Matrigel was removed using cold Cell Recovery Solution under gentle shaking. Whole organoids were fixed for 2 hours using 10% neutral buffered formalin solution and permeabilized with 2% Triton solution in phosphate-buffered saline for 15 minutes. Antibody and staining solutions included DAPI, Phalloidin-647 (A22287), MIST1 (14896S), KI67 (9449S), E-Cadherin (AF748), Alexa Fluor 568 donkey anti-rabbit IgG (A10042), and Alexa Fluor 488 donkey anti-mouse IgG (A21202). 

Paraffin sections of lacrimal gland tissue were deparaffinized using subsequent xylene and ethanol washes. A 10-minute, 125°C incubation in citrate buffer (1 g NaOH, 2.1 g citric acid, 1 L H₂0, pH 6.0) was used for antigen retrieval. Slides were washed and blocked for 45 minutes using DAKO serum-free protein block. 

For RNA in situ hybridization, the lacrimal gland tissue was fixed in 10% neutral buffered formalin solution. Paraffin sections were processed for chromogenic or fluorescent RNA in situ hybridization following the Advanced Cell Diagnostics (Newark, CA, USA) RNAscope reagents and protocols. 

RNA Isolation and Quantitative Polymerase Chain Reaction

Both lacrimal gland organoids and primary tissue were lysed and processed for RNA extraction following manufacturers instruction (Qiagen RNeasy Plus Mini Kit; Qiagen, Hilden, Germany). The digestion of tissue was enhanced further using shaking with metal beads in the RLT lysis buffer. The quality and quantity of the RNA was checked before complementary DNA (cDNA) synthesis using a VILO cDNA synthesis kit. The quantitative polymerase chain reaction was performed using SYBR green on a Bio-Rad machine and analyzed using CFX software. Primers included: Actb_F: ACCTTCTACAATGAGCTGCGT, Actb_R: AGGTCTCAAACATGATCTGGGT, Axin2_F: CAGCCCAAGAACCGGGAAAT, Axin2_R: GAGCCTCCTCTCTTTTACAGCA, Krt7_F: GCGGAGATGAACCGCTCTAT, Krt7_R: TCTAACTTGGCACGCTGGTT, Bhlha15_F: CCAAGATCGAGACCCTCACG, and Bhlha15_R: GCGGCTGCTGGACATAGTAA. 

Duct Ligation Model

Lacrimal gland duct ligation was performed as previously described (Liu et al.³⁷). Briefly, the animal is operated under a dissecting scope fashioned with a nosecone to continue anesthesia by isoflurane. A small incision anterior to the ear is made to expose the extraorbital lacrimal gland which is lifted to expose the main excretory duct. A size 4-0 silk thread is then used to ligate the duct. After duct ligation, the skin incision is closed, sutured, and antibiotic ointment is then applied to the skin. The contralateral lacrimal gland is not operated on and served as a control. After 3 days, the skin suture is reopened to expose the lacrimal gland. The duct ligation is then released by cutting the silk suture with micro scissors. In the same surgery, treatment groups are injected with 2 µL of test article into the lacrimal gland ipsilateral to the ligation. After injection, the skin incision is closed, sutured, and antibiotic ointment is then applied. The animals are then sacrificed, and ipsilateral and contralateral lacrimal glands collected 1 or 3 days after releasing the ligation. Lacrimal glands were weighed immediately after isolation on an analytical balance. The weight of the gland was normalized to the body weight on the day or termination. A minimum of five glands from either side was measured per time point. 

Single-Cell RNA Sequencing (scRNA-seq) Analysis

Cell Handling, Sequencing, Alignment, Filtering, Normalization

One lacrimal gland from each mouse was dissociated into a single cell suspension. Live cells were selected via DAPI exclusion using a Sony SH800 cell sorter. After collection, cells were centrifuged (400 rcf) and resuspended at a higher concentration. We used 10x Genomics 3′ v3.1 reagents for cell capture and library preparation, and approximately 4000 to 5000 cells per sample were loaded for each sample. Multiplexed cDNA libraries were sequenced on one lane of a Nova Seq 6000. Alignment and demultiplexing were done using the 10x Genomics Cellranger pipeline.²² R was used for all subsequent analyses. 

Although we attempted to collect samples at day 3, there was a clog in the microfluidic capture device resulting in extremely low-quality samples, and these were excluded bioinformatically. After exploring the data, we applied the following filtering criteria: only cells with less than or equal to a mitochondrial gene percentage of 17 (a little less than the mean +1 SD), at least 1000 but no more than 65,000 Unique Molecular Identifiers, at least 500 but no more than 7500 genes detected. Only genes that were in the upper quartile of expression in at least three cells were retained. These filters ultimately resulted in a dataset of 17,079 genes and 43,197 cells. Scaling normalization by deconvolution of size factors was applied to the dataset using the scran (version 1.30.2)²³ and scuttle (version 1.12.0)²⁴ R packages. 

Clustering, Differential Expression, and Pathway Analysis

The full dataset was clustered based on the first 10 principal components of the normalized data using shared nearest neighbor graph-based clustering with the buildSNNgraph function from the scran (1.30.2) package and the cluster_louvain function (k = 60) from the igraph (2.0.3) package.²⁵ This strategy yielded 20 clusters, which allowed us to discriminate cells that comprise the lacrimal gland proper, including epithelial cells and associated stroma and endothelial cells and the immune cell lineage. At this point, the dataset was subsetted into two smaller datasets, the lacrimal gland with associated stromal and endothelial cells (25,089 cells) and the immune lineage (18,108 cells). We included in the immune lineage some very small groups of cells that coexpressed either macrophage or neutrophil markers along with markers of specific lacrimal gland cell types. Future endeavors could try to validate whether these cells represent phagocytic cells that have engulfed damaged lacrimal gland cells and debris. 

A negative binomial regression model was applied to the top 2000 most variable genes of the lacrimal gland proper (including associated stromal and endothelial cells) dataset using the newFit function from the NewWave (1.12.0) R package²⁶ with K set to 10. The 10-dimension matrix of inferred sample-level covariates was used as input for clustering the lacrimal gland dataset with the buildSNNgraph (k = 85) and cluster_walktrap functions from the R packages, scran (1.30.2) and igraph (2.0.3), respectively. This process yielded 22 clusters, which were annotated using established cell type markers. A similar approach was used for the immune lineage: after an initial clustering, the lymphoid and myeloid lineages were clustered separately and annotated using established cell-type markers. 

Cluster or cell-type–enriched markers were identified by applying differential expression analysis between clusters at the cell level using limma (3.58.1)²⁷ with a wrapper function from the clusterExperiment (2.22.0) package²⁸ or at the aggregated level (aggregated by batch) using edgeR (4.0.16).²⁹ Differential expression between treatment groups within cell types was done at the batch-aggregated level with edgeR (4.0.16). Gene set enrichment analysis was done with the goanna, kegga, and fry functions from limma. CellChat (2.1.2)³⁰ was used for plotting predicted cell-to-cell communication networks. 

Code and Data Availability

The FASTQ files, a normalized counts matrix, and metadata including experimental conditions and cell type annotation have been deposited at GEO, as record GSE295860. Code for the full analysis is available upon request. 

Results

WNT Mimetic Activation Enables Acinar Cell Organoid Expansion

To investigate the role of WNT signaling in the lacrimal gland, we started with naïve adult murine tissue. Freshly isolated lacrimal glands were fixed for WNT receptor profiling (Fzd_1–10, Lrp5, and Lrp6) through RNA in situ hybridization or were dissociated and plated for organoid outgrowth from primary cells (Fig. 1A). Among the 10 Frizzled receptors, Fzd₅ and Fzd₇ mRNA were most abundant within the gland (Fig. 1B). The other co-receptors essential for functional signaling, Lrp5 and Lrp6, were also detected in the glands (Supplementary Fig. S1A). To test whether cells might functionally transduce signaling through these receptors, we incubated freshly isolated primary cells with a combination of the broad specific WNT mimetic L-F12578 (targeting LRP5, LRP6, and FZD_1,2,5,7,8) and the pathway enhancer RSPO.^21,31,32 For cell outgrowth, we applied a three-dimensional adult stem cell–based organoid protocol frequently used for other organs and lacrimal gland ductal organoids.¹⁰ Wild-type murine lacrimal glands were divided into smaller pieces, incubated in digestion buffer, and embedded in Matrigel basement membrane matrix. Organoid medium was supplemented with B27, N2, EGF, FGF2, and FGF10 in the absence or presence of L-F12578 and/or RSPO. Within 1 week of primary cell plating, all conditions showed three-dimensional outgrowth of cell clusters, albeit with differing morphologies. Basal organoid medium without any WNT signaling modulators, or with RSPO alone, showed a cystic phenotype similar to that previously described for murine ductal organoids (Fig. 1C).¹⁰ Primary cells incubated with L-F12578 or a combination of L-F12578 and RSPO grew into organoid-like structures that were much denser than ductal organoids and showed more budding (Fig. 1C). The addition of EGF and FGF in combination with WNT pathway activation appeared to be essential to achieving this outgrowth (Supplementary Fig. S1B). A model of WNT signaling in combination with other growth factors to provide both stemness and mitotic signals has been described in the mammary gland.³³ The budding phenotype closely resembles other glandular organoids previously described for mammary glands and salivary glands.^34,35 These WNT-dependent organoids were expanded successfully through multiple passages by using mechanical shearing and occasionally displayed budding elongation with a clear lumen (Fig. 1C and Supplementary Fig. S1C). Expression profiling of the various phenotypes revealed that WNT-responsive lacrimal gland organoids had significantly higher expression of the WNT target gene Axin2 and acinar cell marker Bhlha15/Mist1 in the presence rather than the absence of L-F12578 (Fig. 1D). As expected, the combination of L-F12578 and RSPO further elevates Axin2 compared with L-F12578 alone. The expression of Mist1 correlated with a lower expression of ductal cell marker Krt7 (Fig. 1D). We confirmed that the new organoids consisted of a polarized epithelium expressing the acinar cell marker MIST1 protein (Fig. 1E). Many organoid cells in high WNT expansion medium showed double positivity for MIST1 and KI67, thus confirming their proliferative and expanding acinar cell state (Fig. 1E). Occasionally, MIST1-negative organoids were detected in the cultures. These structures were morphologically different from the rest of the culture and were able to be picked manually and removed (Supplementary Fig. S1D). To confirm the epithelial origin of these cultures, we isolated and plated single EPCAM⁺ cells from primary lacrimal gland tissue. Within 10 days, single cells grew into MIST1-expressing budding phenotype organoids (Fig. 1F). We called organoids with this new phenotype formed under WNT mimetic expansion medium lacrimal gland acinar cell organoids. 

Figure 1.

 

WNT activation is essential for acinar cell organoid expansion. (A) Illustration of experimental setup. Murine lacrimal glands are used for receptor expression profiling or processed for organoid culture. (B) Representative images of HE staining, and Fzd5 and Fzd7 RNA in situ hybridization. The signal is in pink. Scale bars, 100 µm. (C) One-week outgrowth of primary cells with a three-dimensional organoid protocol. Representative images of control treatment, and RSPO2 treatment with or without L-F12578. (Right) Close-up of passage 7 budding organoid expanded in RSPO2 and L-F12578. (Left) Scale bars, 1000 µm. (Close-up) Scale bar, 100 µm. (D) Gene expression quantification through quantitative polymerase chain reaction of the WNT target gene Axin2 and the acinar cell marker Bhlha15/Mist1. (E) Representative images of organoid immunofluorescence staining of nuclei (DAPI), as well as markers of proliferation (MKI67), the cytoskeleton (F-actin), and acinar cells (MIST1). Lacrimal gland tissue image is shown at bottom right as a reference. Scale bars, 25 µm. (F) Representative brightfield images of lacrimal gland organoid outgrowth over the course of 10 days from single EPCAM⁺ plated cells. Scale bars, 100 µm.

FZD Subtype-Specific Mimetics Reveal the Importance of FZD_1,2,7

We previously explored the effects of recombinant WNT mimetics targeting all eight WNT/β-catenin signaling FZD receptors in several other tissues.^21,36 To elucidate the fundamental WNT receptors in the adult lacrimal gland, we applied a similar approach using our WNT mimetic toolbox.²¹ Starting in vitro, we tested L-F12578, L-F127, L-F58, L-F4, L-F49, and L-F10 according to our lacrimal gland adult stem cell–based organoid protocol. To prevent any endogenous WNT protein secretion and signaling, we added the porcupine inhibitor C59. As expected, outgrowth with L-F12578 resulted in an acinar cell organoid phenotype, as did the more subfamily specific WNT mimetics L-F127 and L-F58 (Fig. 2A and Supplementary Fig. S2A). Activation through FZD₄, FZD₉, or FZD₁₀ by using L-F4, L-F49, and L-F10 resulted in organoid outgrowth, but led to a phenotype resembling the ductal phenotype observed under the control condition (Fig. 2A and Supplementary Fig. S2A).¹⁰ The difference in outgrowth was also observed in cell viability assays measuring adenosine triphosphate levels: WNT mimetics targeted to FZD_1,2,7 and/or FZD_5,8 elicited the largest increase in detectable viable cells (Fig. 2B). Whereas L-F4 stimulation elicited a significantly greater increase in viable cells than the control, the phenotype markedly differed from that associated with the other WNT mimetics. To further understand the difference in receptor signal transduction, we examined short-term direct target gene induction. After 24 hours or 48 hours incubation of established organoid cells with WNT mimetics, a robust increase in the expression of the target gene Axin2 was observed, and the L-F127 WNT mimetic elicited the most pronounced Axin2 elevation at the 48 hours time point (Fig. 2C). Although Axin2 induction with L-F58 was not statistically significant, it was enough to induce acinar cell proliferation. Lacrimal gland epithelial cell activation through FZD_1,2,7 by using L-F127 stimulated organoid formation in a dose-dependent manner, with doses as low as 10 pM (Figs. 2D and 2E and Supplementary Fig. S2B). 

Figure 2.

 

WNT mimetic targeting FZD1, 2, and 7 stimulates acinar cell proliferation in vitro and in vivo. (A) Representative brightfield images of organoids expanded for 4 days from primary lacrimal gland cells. Treatments included WNT mimetic with FZD subfamily specific targeting, or no-WNT control. Scale bars, 100 µm. (B) Cell viability quantification of adenosine triphosphate (ATP) levels in organoid outgrowth, visualized in (A). Data were normalized to the control. (C) Acinar cell organoid gene expression changes, quantified with quantitative polymerase chain reaction, after 24 hours or 48 hours incubation with FZD subfamily specific WNT mimetics. Data were normalized to the no-WNT control. (D) Organoid expansion (6 days), measured according to cell viability ATP levels. Treatments included a range of FZD127 WNT mimetic (L-F127) doses. Data were normalized to the no-WNT control. (E) Representative images of organoid outgrowth (6 days) quantified in (D); control and 10 nM WNT mimetic. The total image captures a 20 µL Matrigel droplet with cells and organoids. (F) Schematic illustration of the animal dosing scheme. A total of 24 animals (n = 12/group) were administered either vehicle control or L-F127 WNT mimetic for 2 weeks. (G) Representative images of immunofluorescence staining for proliferation in adult lacrimal glands after 2 weeks, with staining for nuclei (DAPI, blue) and the proliferation marker MKI67 (magenta). (Right) Representative H&E staining images of a lacrimal gland in each treatment group. Scale bars, 25 µm in immunofluorescent stain and 100 µm in H&E stain.

To test WNT mimetic effects in vivo, we administered L-F127 or vehicle control intraperitoneally at 10 mg per kg on days 0, 3, 7, and 10 to naïve mice (Fig. 2F). After 2 weeks of treatment, we fixed the lacrimal glands and stained sections for MKI67 to assess pro-proliferative effects, as observed in vitro. On day 14, the L-F127–treated group showed detectable KI67-postive cells in the lacrimal gland, whereas the vehicle control group showed few to no signs of proliferation, as would be expected in healthy glands (Fig. 2G). Visualization by DAPI staining indicated that these proliferative cells had large nuclei and therefore were likely to be acinar cells. Compared with lacrimal glands in vehicle-treated animals, L-F127–treated lacrimal glands had a slightly different appearance in histological sections. After hematoxylin and eosin¹⁰ staining, the L-F127 treated gland acinar compartment seemed to be larger, with an increase in basophilic cytoplasm compared with control (Fig. 2G and Supplementary Fig. S2C). We have observed a similar phenotype in murine salivary glands activated by WNT mimetic treatment in vivo.²¹ 

L-F127 Supports Lacrimal Gland Regeneration in Duct Ligation Damage

To investigate whether the pro-proliferative effect observed with L-F127 WNT mimetic treatment might be used to stimulate lacrimal gland regeneration, we used a murine dry eye model allowing for local damage and local administration of WNT mimetic. Ligation of the main excretory duct of the lacrimal gland is an established model for tissue injury, inflammation, and diminished tear volume.³⁷ With small adjustments in the protocol, we sought to establish the timeline of endogenous recovery and determine the treatment window for this model. For each mouse, on one side, we ligated the main duct for 3 days (ipsilateral side) and used the other side (contralateral side) as a control for both tissue assessment and tear volume measurements. After 3 days, we removed the suture and reopened the duct (Fig. 3A). Histological examination of the glands immediately after 3-day ligation showed significant shrinkage of the epithelial compartment, with atrophic acinar cells and immune cell infiltration (Fig. 3B). Consequently, the ipsilateral glands were generally smaller than the contralateral glands. This injury phenotype persisted around day 7 but began to return to normal on day 14 (Fig. 3B). By day 21, the ligated lacrimal glands seemed to have recovered and showed a similar histological phenotype to that of the contralateral control glands (Fig. 3B). Tracking tear fluid secretion during these 3 weeks with a phenol red thread applied to the ocular surface revealed a strong decrease in tear volume after injury (Fig. 3C). Similar to the tissue recovery observed in histological sections, the tear volumes returned to contralateral levels in the last 2 weeks of the time course (Fig. 3C). The lacrimal gland tissue weight decreased 7 days and 14 days after injury, then returned to a weight similar to that on the contralateral control side by day 21 (Fig. 3D). A compensatory mechanism may exist that enables the lacrimal gland to secrete normal volumes by day 14 despite a reduced organ weight. One of the animals' glands did not show weight recovery and remained small for 3 weeks. We hypothesized that the injury might occasionally be too severe to enable recovery. Together, our findings indicated endogenous recovery of the murine lacrimal gland after duct ligation injury. We concluded that any beneficial effects of WNT mimetic treatment are to be observed within a 2-week treatment window before the glands return to normal. 

Figure 3.

 

L-F127 WNT mimetic accelerates recovery in a duct ligation model of dry eye. (A) Schematic illustration of duct ligation injury in the adult murine gland; a 3-day closure of a major excretory duct (ipsilateral side, red) and one undamaged control side (contralateral side, blue) is shown. (B) Representative H&E images of lacrimal glands at different time points after 3-day duct ligation, compared with a contralateral control. (Top right) Changes in gland morphology after a 3-day ligation. Scale bar, 1 mm. (C) Quantification of tear volumes with phenol red thread (PRT), measured on different days on either the contralateral side (blue) or damaged ipsilateral side (16). (D) Measurements of lacrimal gland (22) weight, normalized to body weight (BW), on the ipsilateral and contralateral sides on day 7, 14, and 21. Each datapoint represents a single gland from one animal. (E) Visualization of Fzd7 mRNA signal (pink) via in situ hybridization in lacrimal gland tissue 14 days after a 3-day duct ligation. (Bottom) Close-up of the boxed area in top. Scale bar, 100 µm. (F) Schematic illustration of the study design used to test local treatment with L-F127 WNT mimetic in the duct ligation model of dry eye. (G) Gene expression changes in duct ligation injured lacrimal gland tissue 24 hours after local treatment with two doses of L-F127. Expression of the WNT target gene Axin2, normalized to that in duct ligation control treated glands. (H) Quantification of tear volumes with phenol red thread (PRT), measured on different days on the ipsilateral side in control treated glands or L-F127 treated glands at two dose levels.

To confirm that WNT receptors remain expressed after duct ligation, we probed for Fzd₇ mRNA by using in situ hybridization. The expression of Fzd₇ in injured tissue showed a similar pattern to that observed in naïve lacrimal glands. Therefore, we next tested the administration of L-F127 after a 3-day duct ligation. To develop L-F127 into a potential therapeutic candidate, we humanized both anti-LRP5 and anti-FZD_1,2,7 domains and removed potential chemical instability sites. We named this final variant L-F127-v2. We established a small volume local injection of L-F127-v2 or negative control during the same surgery to remove the suture on day 3. Local administration enabled control of the active dose levels and potentially diminished systemic adverse effects. With this design, we collected tissues 1, 7, and 14 days after ligation treatment (Fig. 3F). We used the early time point to confirm WNT pathway activation after treatment. Compared with vehicle control–treated glands, the 10 µg and 100 µg L-F127-v2 dose levels both significantly elevated expression of the WNT target gene Axin2 at 24 hours (Fig. 3G). Local WNT pathway activation was followed by an increase in tear volume starting from day 7 (Fig. 3H). Both dose levels of L-F127-v2 accelerated tear secretion recovery, and the phenol red thread length was significantly greater on days 7, 12, and 14 than observed after vehicle control treatment (Fig. 3H). In an independent study that was terminated on day 7, we observed a similar significant treatment-induced increase in tear volume (Supplementary Fig. S3B). We did not detect any differences in general histology between treatment groups (Supplementary Fig. S3C). 

L-F127-v2 Affects Injury-Induced Cell States

To gain a mechanistic understanding of how L-F127-v2 affects the injured lacrimal gland and enhances tear volume recovery, we applied scRNA-seq after duct ligation injury. Cells were isolated and sequenced from uninjured, untreated lacrimal glands, as well as from injured lacrimal glands after 3 days of ligation (day 0). We also analyzed cells from injured glands on days 1, 5, and 7 after duct ligation removal and treatment, comparing L-F127-v2 with a control antibody (Figs. 4A and S4A). After filtering, the dataset included 43,197 cells; after normalization, the cells were segregated by tissue layer and lineage (Supplementary Figs. S4B and S4C). We detected more than 40 cell types and states, including the cell types identified in previous scRNA-seq of the adult murine lacrimal gland, such as Car6⁺ acinar/ductal cells^38,39 (Fig. 4B and Supplementary Figs. S4C and S4D). 

Figure 4.

 

Duct ligation severely damages acinar cells and induces injury-specific epithelial cell states. (A) The 3-day duct ligation injury model design for scRNA-seq. (B) Dot plot of select cell type marker expression for all lacrimal gland cell types/states identified. (C, D) Reduced dimension plot (UMAP) of the lacrimal gland, excluding immune cells, colored by cell type (C) or experimental condition (D). (E) Proportion of each cell type within a given experimental condition, displayed as a percentage summing to 100% for each condition (column). (F) Gene expression for two established secretory/acinar cell markers, Bhlha15/Mist1 and Sval2. (G) Immunofluorescence of BHLHA15 (MIST1) and CDH1 (ECAD) on an uninjured, contralateral gland or injured gland at 7 days after duct ligation.

We next focused on the lacrimal gland epithelial, stromal, and endothelial cells and observed that the duct ligation injury severely affected the acinar cells, which were undetectable at days 0 and 1 after the 3-day duct ligation (Figs. 4C and 4E). In fact, mature, fully differentiated acinar cells did not return during the 7-day time course, although newly formed acinar cells showing increasing expression of differentiation markers were present at days 5 and 7, and clustered with the uninjured acinar cells (Figs. 4C–F and Supplementary Fig. S4E). This finding was validated in tissues, in which the number of acinar cells detected by BHLHA15/MIST1 remained diminished 7 days after injury (Fig. 4G). By comparing WNT receptor expression levels across the cell populations, we found Fzd7 to be enriched in epithelial cells and Fzd4 in endothelial cells, whereas the co-receptors Lrp5 and Lrp6 were more broadly expressed across all lacrimal gland cell types (Supplementary Fig. S5B). 

The duct ligation injury led to the formation of injury-specific cell states that were transient and did not persist through day 7. One cell type/state derived from injured samples was a basal ductal cell state (BasalDuctal_Inj), populated predominantly by injury-induced cells through day 5. We further observed two very small groups of acinar-like cells that were challenging to define, because they had low expression of most cell type marker genes, except for some secretory genes such as Esp15 (Fig. 4B), they existed transiently (Fig. 5A), and they did not respond to L-F127-v2 treatment in differential expression analysis. Moreover, three clusters of injury-induced cells showed high expression of luminal ductal genes (Fig. 4B and Supplementary Fig. S4D), and two groups also showed relatively high expression of Car6; we termed these cells LumDuct_Inj and C6LumDuct_Inj1 and 2 (Figs. 4B and 5A). These cells might potentially have originated from luminal ductal and Car6⁺ cells, because of the similarity in gene expression, and because normal luminal ductal and Car6⁺ cells^38,39 were not present after injury at day 0 or 1 (Fig. 5A), when these injury-induced states are predominant. However, lineage tracing would be necessary to provide confirmation of this possibility. 

Figure 5.

 

L-F127 WNT mimetic directly affects injury-induced Car6⁺ and ductal epithelial cell states. (A) UMAP of the lacrimal gland proper (minus immune cells), colored by cell type/state or experimental condition by injury time point. Injury-induced cell states are circled in black (solid or dashed); cells showing a transcriptional response to L-F127 are circled in red (top left UMAP). A legend for all cell types/states and conditions is shown at left. (B, C) Dot plots of key direct WNT target (Axin2, Lgr5, and Znrf3) or cell-cycle–associated gene expression that significantly increased after L-F127 treatment at day 1 in the injury-induced Car6⁺ luminal ductal cell state (B) and the luminal ductal cell state (C). (D) Quantification of the percentage of acinar cells (BHLHA15/MIST1 positive) that were MKI67 positive at day 14 after duct ligation injury after the indicated treatment. (E) Example images of injured lacrimal gland tissue at day 14 after the indicated treatment.

Compared with the control treatment, L-F127-v2 treatment significantly upregulated the expression of established WNT signaling target genes (Axin2 and Lgr5) within two of these injury-induced cell states (LumDuct_Inj and C6LumDuct_Inj1) at day 1 (24 hours after treatment), in agreement with these cells being proximal cells responding to the L-F127-v2 WNT mimetic (Figs. 5A [red oval]–C, and Supplementary Table S1). In addition, L-F127-v2 treatment increased cell-cycle gene expression at the whole gland level (Supplementary Fig. S5C) and specifically in these two injury-induced cell states at day 1 (Figs. 5B and 5C, Supplementary Fig. S5D, and Supplementary Table S1). These responses were captured in gene set enrichment analysis, wherein L-F127-v2 led to enrichment in pathways associated with WNT signaling, gland development, and proliferation (Supplementary Figs. S5E and S5F and Supplementary Table S1) in the LumDuct_Inj and C6LumDuct_Inj1 cells. These findings suggested that L-F127-v2 treatment affected injury-induced Car6⁺ and luminal ductal cells and enhanced the initiation of a proliferative response to injury. In agreement with this earlier effect, by day 5, we observed very few Car6⁺ and luminal ductal-like injury induced cells (Fig. 5A), as well as repopulation of more normal Car6⁺ and luminal ductal cells, with a relatively high contribution from the L-F127-v2 treated cells (Fig. 4E and Supplementary Fig. S4E). At 14 days after duct ligation, we still detected a treatment induced increase in proliferation. At this time point, Mist1-expressing acinar cells had repopulated the gland. Approximately 12% of acinar cells were Mist1⁺/Ki67⁺, thus indicating a subset of proliferative acinar cells (Figs. 5D and 5E). Whereas the control treated group showed less proliferation than both treatment groups, it remained around 10% Mist1⁺/Ki67⁺, as part of the endogenous response to damage. 

Differential expression analysis identified significant enrichment in Axin2, a key WNT target gene, at day 5 in three cell types (Myoepithelial1, LuminalDuctal1, and BasalDuctal2), as well as a trend toward enrichment in LuminalDuctal2 (false discovery rate, 0.08) (dashed circles in Fig. 5A; Supplementary Table S1) under L-F127-v2 compared with control treatment. Critically, at day 5, the WNT mimetic L-F127-v2 would not be expected to cause direct pharmacodynamic effects, because of its serum half-life. One interpretation of the findings might be that Axin2 expression persisted from initial pathway activation at day 1 in the injury-induced states (LumDuctal_Inj and C6LumDuctal_Inj1) that differentiated into these cell types. Future work could address the lineage relationships of the injury-induced cell types/states. Although the fibroblasts showed almost no treatment response, Lgr5 was significantly elevated in Fibroblast1 cells. 

We observed injury-induced and injury-enriched immune cell types, but no evidence of a direct WNT signaling pathway activation in any immune cells. For example, several subgroups of macrophages were either specific to injury or highly enriched under the injury condition (Supplementary Figs. S6A, S6B, and S6D–S6F). Neutrophils were present only at the end of the duct ligation (day 0) and persisted through day 7, when a strong enrichment in neutrophils from days 5 and 7 was observed under L-F127-v2 versus control treatment (Supplementary Figs. S6D–S6G). Examination of predicted cell-to-cell communication indicated that the injury-induced cell types/states and the Car6⁺ and luminal ductal cells expressed several chemokines that would be expected to recruit neutrophils. 

Discussion

The WNT/β-catenin signaling pathway is a major regulator of tissue homeostasis and stem cell renewal in adults.⁷ Natural ligands are hydrophobic, poorly expressed, and lack defined receptor specificity.^40–42 The discovery of soluble WNT mimetics targeting various FZDs circumvented these challenges and enabled a platform for targeted regeneration.²⁰ Herein, we described the use of our WNT-modulating platform to better understand adult lacrimal gland biology. We additionally propose a new therapeutic approach for regenerative medicine. 

Recent advances in three-dimensional cell culture technology have enabled adult mammalian stem cells to be expanded as organoids.⁴³ These self-organizing structures recapitulate their tissue of origin and can be used to model signaling pathways during homeostasis and disease.⁴⁴ Adult stem cell-derived organoids can be generated from several human and mouse organs, most successfully from epithelial cell types. In many organoid protocols, WNT signaling through WNT3A or RSPO plays an essential role in the activation and expansion of cells.⁴⁵ Pathway activation with a WNT mimetic is essential for expanding lacrimal gland acinar cells in vitro. The expression of Fzd₇ was detectable in tissue, and signaling through FZD₇ by using an LRP-FZD127 WNT mimetic was found to be most potent. A combination of WNT mimetic and EGF and FGF activation allowed for lacrimal gland acinar cell organoid expansion and passaging. Previous attempts without the use of WNT mimetic and EGF addition enabled the expansion of ductal cells.¹⁰ Other differences with respect to the ductal medium include the lack of Noggin, A83-01, Forskolin, and prostaglandin E2 in acinar cell medium, a more minimal expansion medium. The possibility of trans differentiation between these cell states in vitro and the exact cell of origin in both cultures warrants further investigation. Because of challenges in obtaining high-quality primary human lacrimal gland tissue, we were unable to test the effects of WNT mimetics in the generation of human organoids. 

The effects of L-F127 were further investigated in vivo. Although no animal model exists that mirrors all aspects of human disease, we believe that the duct ligation model recapitulates both the epithelial damage and diminished tear volume. In this mouse model of dry eye disease, we tested the effects of local administration of L-F127 on tissue recovery, acinar cell proliferation, and tear volume output. Our results revealed that WNT mimetic treatment activated the pathway in an injured environment after a 3-day duct ligation. This pathway activation was followed by an increase in tear volume starting 1 week after treatment. We hypothesized that acute local WNT activation initiated transcriptional changes in target cells that resulted in improved functional output and proliferation several days later. A local injection might also be considered for future treatment in humans, because the glands are readily accessible.⁴⁶ 

Single-cell mRNA sequencing technology has become a powerful tool to reveal the cellular composition of a tissue and discover changes during injury or disease.⁴⁷ Although other researchers have generated single-cell datasets of normal murine lacrimal glands and those with genetic Sjogren's syndrome–like phenotypes, no data are available for the duct ligation model or WNT mimetic treatment.^38,39 We applied scRNA-seq to murine lacrimal glands after duct ligation injury, and subsequent L-F127-v2 or control treatment, or to uninjured glands. We thus identified (1) the cellular composition changes that occur after injury and (2) the effects of WNT mimetic treatment on these injured cell states. First, the effects of a 3-day duct ligation were found to be broad and severe, particularly in the epithelial compartment. Mature, fully differentiated acinar cells were not detectable in the first week after injury. This finding was consistent with the striking atrophy visible in histological sections. Transient cell types and states specifically derived from injured samples included basal ductal, luminal ductal, and acinar like. Lineage plasticity and the ability of postnatal epithelial cell types to transdifferentiate have been described in the lacrimal gland.⁴⁸ Consequently, the gland's cellular composition and cell states present at the start of treatment differed from those observed in the profiling of healthy glands or the acinar cells isolated to initiate organoid growth. 

Two injury-induced luminal ductal cell states, denoted LumDuct_Inj and C6LumDuct_Inj1, showed significant expression changes on day 1 after treatment. Local injection of L-F127-v2 resulted in the upregulation of the established WNT target genes Axin2 and Lgr5, as well as an increase in cell-cycle signature genes 24 hours after treatment. At later time points, we detected Axin2 elevation in a different subset of cell states and an increase in acinar cell proliferation in L-F127-v2–treated glands. Recent studies have identified Car6⁺ cells as a previously undescribed epithelial cell type sharing features with both acinar and ductal cells.^38,39 These cells also reside at the junction between acini and ducts and have been hypothesized to be a progenitor pool of cells for tissue repair.^38,39 These cells also seemed to be activated by L-F127-v2 WNT mimetic treatment. Fully mature acinar cells were lost during damage and reemerged after several days from progenitor cells. Whether these progenitor cells arose from damaged acinar cells or transdifferentiated from the duct, and the effects of WNT activation on this process, remains to be determined. Given the severity of the damage, linking these transcriptional changes to the functional improvements in tear volume observed starting at day 7 is challenging. Changes in the secretory machinery at the protein level, myoepithelial cell contraction, and nerve innervation will not be detected in this transcriptome dataset. Lacrimal gland acinar cell organoids might be used to study the effects of WNT signaling on secretion mechanisms. WNT activation might potentially lead to a broader tissue remodeling cascade involving multiple cell types. 

Together, our results demonstrated that a full antibody-based WNT mimetic can be used to better understand the biology of the lacrimal gland and may serve as a therapeutic candidate for lacrimal regeneration in dry eye disease. 

Study Limitations

The use of mouse models to predict human responses carries risks, particularly regarding the complexity of underlying disease phenotypes. This study focused solely on mouse organoids and a mouse model of dry eye disease. To demonstrate the potential of WNT signaling activation in lacrimal gland regeneration fully, additional human models should be explored. In addition, we relied on female mouse lacrimal glands in our dataset. Sex differences can be significant, including in Sjogren's syndrome. 

Finally, we did not explore the safety or potential adverse effects of this platform. Although we have some understanding of WNT-responsive tissues, a profile for this particular molecule should be established. Finally, whereas local lacrimal gland injections are performed in humans, the effects of WNT mimetic treatment on other ocular cell types, and of systemic exposure, should be further explored. Future studies should aim to validate the effectiveness and safety of L-F127-v2 in human clinical trials. 

Acknowledgments

The authors thank all current and former colleagues at Surrozen for technical support and helpful discussions. Special thanks to Sungjin Lee, Haili Zhang, Sean Bell, and Yiran Yang for their leading role and support in the development of L-F127(-v2). We thank Roel Nusse, Calvin Kuo, Christopher Garcia, Willard H. Dere, Harold Varmus, Bart Williams, and Hans Clevers for helpful discussions and guidance. All research funded by Surrozen, Inc. No external funding. 

Author Contributions: H.N., Conceptualization, Formal analysis, Supervision, Investigation, Visualization, Methodology, Writing review and editing. R.F., Single cell RNA sequencing: Conceptualization, Formal analysis, Supervision, Investigation, Visualization, Methodology, Writing. T.L., with T.Z.Y., H.G., J.Y., P.S., N.S., H.C., Protein science (L-F127): Conceptualization, Formal analysis, Supervision, Resources, Investigation, Visualization, Methodology. E.W., K.L., N.D., T.S., Mouse model and in vitro assay: Conceptualization, Formal analysis, Supervision, Resources, Investigation, Visualization, Methodology. W.C.Y., Resources, Supervision, Funding acquisition. Y.L., Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing review and editing. Y.P. Conceptualization, Formal analysis, Resources, Supervision, Visualization, Methodology, Writing. 

Declaration of Interests: All authors are current or former full-time employees and shareholders of Surrozen, Inc. Y.L. is Executive Vice President of Research at Surrozen, Inc. Patent applications are pending for the work described herein. 

Disclosure: H. Nguyen, Surrozen, Inc. (E, F); R.B. Fletcher, Surrozen, Inc. (E, F); T. Lopez, Surrozen, Inc. (E, F); E. Whisler, Surrozen, Inc. (E, F); K.R. Logas, Surrozen, Inc. (E, F); N. Dhaliwal, Surrozen, Inc. (E, F); T. Suen, Surrozen, Inc. (E, F); M. Zhang, Surrozen, Inc. (E, F); A. Dilip, Surrozen, Inc. (E, F); T.Z. Yuan, Surrozen, Inc. (E, F); H. Go, Surrozen, Inc. (E, F); J. Ye, Surrozen, Inc. (E, F); P. Sampathkumar, Surrozen, Inc. (E, F); N. Suen, Surrozen, Inc. (E, F); H. Chen, Surrozen, Inc. (E, F); W.-C. Yeh, Surrozen, Inc. (E, P, F); Y. Li, Surrozen, Inc. (E, P, F); Y. Post, Surrozen, Inc. (E, F)