Mass cytometry and transcriptomic profiling reveal body‐wide pathology induced by Loxl1 deficiency

Abstract Objective The loss of LOXL1 expression reportedly leads to the prolapse of pelvic organs or to exfoliation syndrome glaucoma. Increasing evidence suggests that LOXL1 deficiency is associated with the pathogenesis of several other diseases. However, the characterization of the systemic functions of LOXL1 is limited by the lack of relevant investigative technologies. Materials and Methods To determine the functions of LOXL1, a novel method for body‐wide organ transcriptome profiling, combined with single‐cell mass cytometry, was developed. A body‐wide organ transcriptomic (BOT) map was created by RNA‐Seq of tissues from 17 organs from both Loxl1 knockout (KO) and wild‐type mice. Results The BOT results indicated the systemic upregulation of genes encoding proteins associated with the immune response and proliferation processes in multiple tissues of KO mice, and histological and immune staining confirmed the hyperplasia and infiltration of local immune cells in the tissues of KO mice. Furthermore, mass cytometry analysis of peripheral blood samples revealed systemic immune changes in KO mice. These findings were well correlated with results obtained from cancer databases. Patients with tumours had higher Loxl1 mutation frequencies, and patients with Loxl1‐mutant tumours showed the upregulation of immune processes and cell proliferation and lower survival rates. Conclusion This study provides an effective strategy for the screening of gene functions in multiple organs and also illustrates the important biological roles of LOXL1 in the cells of multiple organs as well as in systemic immunity.

influence the release of inflammatory factors. 5 Therefore, these processes are related to diverse pathological conditions and diseases, including embryonic lethality, 6 inflammatory response and fibrosis, 7 ageing 8 and cancer. 9 Lysyl oxidase like 1 (LOXL1), an enzyme responsible for elastin synthesis and cross-linking, is needed for elastic fibre homeostasis and ECM stability. Reportedly, the loss of LOXL1 function leads to pelvic organ prolapse 10 and exfoliation syndrome glaucoma. 11 Increasing evidence suggests that Loxl1 deficiency contributes to the pathogenesis of several other diseases, such as cancer, 12 13,14 have been used to demonstrate that Loxl1 deficiency leads to abnormal elastin regulation and subsequently leads to pelvic organ prolapse, 15 flabby skin and bullae. 16 However, owing to the expression of elastin ECM in various organs, its biological effects and clinical relevance in other organs have always been ignored.
Owing to the significant advances in high-throughput sequencing technologies, the biological functions of a single gene can now be characterized at the whole-genome and single-cell levels. 17 In this study, we developed an efficient body-wide organ transcriptomic (BOT) profiling method combined with single-cell mass cytometry to evaluate the biological manifestations of abnormal elastin regulation through single-cell mass cytometry and RNA-Seq of tissues from 17 organs, comparing Loxl1(-/-) mice with wild-type mice.

| Animal model and harvesting of organs
Twelve-week-old Loxl1 KO mice and C57BL/6 mice (both bred at the National Resource Center for Mutant Mice, Model Animal Research Center of Nanjing University) were used in this study. Three animals from each of KO and WT mice groups were humanely killed, and tissue samples were collected from 17 organs (skin, abdominal fat, bone, aorta, brain, lung, bladder, kidney, small intestine, liver, rectum, heart, cartilage, tail tendon, skeleton muscle, vagina and spleen) for the subsequent experiments. Four animals from each group were humanely killed, and the peripheral blood samples were pooled into a single sample for single-cell mass cytometry.

| Ethics Approval
All animal procedures in this study were performed using ethically approved protocols, in accordance with guidelines of The Lab of Animal Experiment Ethical Inspection of College of Medicine, Zhejiang University (Reference number: 2015-112).

| Histological examination
The specimens obtained were immediately fixed in 4% neutral buffered paraformaldehyde, dehydrated by subjecting to an alcohol gradient, cleared and embedded in paraffin blocks. Histological sections (7 μm each) were prepared using a microtome, which were subsequently paraffinized with xylene, hydrated by treating with ethanol at decreasing concentrations and then subjected to haematoxylin and eosin (H&E) staining and Weigart's staining. Next, the sections were mounted and observed under a microscope.

| RNA-Seq
RNA-Seq was performed according to a previously described method. 18 Briefly, RNA was extracted from the samples using TRIzol reagent (TAKARA), reverse transcription was performed using SuperScript II Reverse Transcriptase (Invitrogen), double-strand cDNA was isolated using the NEBNext mRNA second strand synthesis kit, double-strand DNA was purified using AMPure XP beads (Beckman Coulter), and the sequencing library was constructed using the Nextera XT kit (Illumina), followed by sequencing on the Illumina X-Ten platform. The RNA-Seq read data were mapped to the reference genome using TopHat and Cufflinks. 19,20 The expression levels were calculated and expressed in counts per million.

| Data analysis for RNA-Seq
Differentially expressed (DE) genes were analysed using DESeq2, and data were selected at P <.05. 21 Gene ontology (GO) enrichment analysis was performed using the DAVID informatics resources (https://david.ncifc rf.gov/). 22

| Immunofluorescence staining
The tissues were fixed in 4% (w/v) paraformaldehyde, dehydrated by treating with an ethanol gradient, embedded in paraffin and cut into sections with a thickness of 7 μm. Immunostaining was performed as follows. The paraffinized tissue sections were rehydrated, and the antigens were retrieved, followed by rinsing three times with PBS.
Next, the sections were treated with blocking solution (1% BSA) for 30 min before overnight treatment with primary antibodies at 4℃. Rabbit anti-mouse antibodies against KI67 (Abcam, ab16667) and CD45 (BD Biosciences, 555 483) and rat anti-mouse antibodies against CD19 (Biolegend, 115 525) and F4/80 (Biolegend, 123 121) were used as primary antibodies to detect cell proliferation and immune cell infiltration. The secondary antibodies goat anti-rabbit Alexa Fluor 488 (Invitrogen, A11008), goat anti-rabbit Alexa Fluor 546 (Invitrogen, A21430-f) and goat anti-rat CY3 (Beyotime Institute of Biotechnology, A0507) and DAPI (Beyotime Institute of Biotechnology, China) were used to visualize the respective primary antibodies and cell nuclei. All procedures were performed according to the manufacturer's instructions.

| Single-cell mass cytometry
Blood samples from four mice from each group were pooled to obtain sufficient cells for reliable mass cytometry. After lysing the erythrocytes using ACK lysis buffer, the samples were washed with FACS buffer and incubated at 4℃ or on ice. The pooled cells were then stained with cisplatin to distinguish between live and dead cells. After blocking at room temperature for 20 min, the cells were stained with a mixture of metal-tagged antibodies targeting surface antigens for 30 min at room temperature (the complete list of antibodies is provided in Table S1). After washing with FACS buffer and fixation for 20 min at room temperature, the cells were washed with Perm buffer and stained using a mixture of intracellular antibodies (Table S1) for 30 min at room temperature. Next, the cells were fixed and stained with a DNA intercalator overnight at 4℃. After washing with Perm buffer, the cells were incubated with barcodes for 30 min at 4℃. After multiple washes with FACS buffer and ultrapure H 2 O, the cells were analysed using a CyTOF mass cytometer. The raw data acquired were uploaded to a Cytobank web server (Cytobank Inc) for further data processing and for gating dead cells and normalization beads.

| Mass cytometry data analysis
Data analysis was performed using the SPADE 23

| Statistical analysis
A two-tailed t test was used to detect differences in histological results between the KO and WT mice groups. Statistical significance was set at P <.05.

| Body-wide pathology induced by Loxl1 deficiency
As LOXL1 is the key enzyme for the synthesis and assembly of elastin, Loxl1 KO mice exhibited phenotypes associated with abnormal elastic fibres, including obvious POP ( Figure 1A) and loose skin ( Figure 1B). POP is characterized by an enlarged perineal body and a bulge in the rectum and vagina. In humans, the molecular mechanism underlying POP is widely considered to be related to the theory of 'imbalance of elastin' 25 ; therefore, Loxl1 KO mice are always used as an animal model for studies on POP. To observe changes in elastic fibres in KO mice, we performed Weigart's staining specific for elastic fibres in vaginal tissues ( Figure 1C) and skin ( Figure 1D). In the KO mice, elastic fibres in the lamina propria of the vagina and in the skin appeared to have a thin and short rod-shaped structure, and the deposition of elastic fibres near the basement membrane decreased in the KO mice model. In contrast, the elastic fibres were polarized and arranged neatly in the vaginal tissues of WT mice, confirming the reduction of synthesis and failure of cross-linking process of elastic fibres in KO mice. Moreover, the spleens of KO mice were longer and had a larger volume than those of WT mice ( Figure 1E).
The body weight of KO mice decreased significantly compared with that of WT mice with increasing age ( Figure 1F). The observation of these additional phenotypic changes suggested that the effects of Loxl1 deficiency were not limited to the reproductive tracts and also affected the spleen and other organs in mice.

| Upregulation of inflammation and hyperplasia in Loxl1-deficient mice revealed using bodywide organ RNA-Seq
To assess the systemic effects of abnormal elastin production on organs and tissues in Loxl1 KO mice, we performed RNA-Seq using tissues collected from 17 organs (heart, liver, spleen, lung, kidney, skin, brain, aorta, abdominal fat, small intestine, bladder, rectum, vagina, skeletal muscle, bone, cartilage and tail tendon) harvested from WT and KO mice, with biological triplicates for each group (Figure 2A).
We first performed correlation analysis of gene expression for all samples from the WT group. The data revealed a high correlation between the expression profiles of similar tissue types. Tissues from organs that are closely associated, such as the rectum and small intestine, cartilage and bone, and skeletal muscle and skin, could be clustered based on gene expression, indicating good repeatability in the sequencing results ( Figure 2B). Principle component analysis was performed. Figure 2C shows the tSNE results of this data, which indicate a clear distinction in the gene expression profiles between the KO and WT mice groups. This indicates the differential characteristics of the two groups ( Figure 2C).
To further compare the DE genes in various tissues in the two groups, we used DESeq2 in R (http://www.R-proje ct.org) for performing downstream analyses ( Figure 2D). The data indicated that the spleen had the highest number of significantly differentially expressed genes (P <.05), followed by the vagina and skeletal muscle, suggesting that Loxl1 deficiency affects the spleen most significantly. As indicated in the heat map generated for DE genes in all tissues ( Figure 2E), most DE genes were unique to each tissue type, which implied that Loxl1 deficiency exerts tissue-specific effects. GO analysis (https://david.ncifc rf.gov/) of the DE genes in different tissues also confirmed the tissue-specific changes following Loxl1 KO. For example, DE genes upregulated in the spleen were enriched with genes associated with the cell cycle and cell division after Loxl1 KO, whereas the downregulated DE genes were those associated with the immune response ( Figure 2F). In contrast, genes encoding proteins related to keratinization and keratinocyte differentiation were upregulated, whereas those encoding proteins related to sarcomere organization and response to estradiol were downregulated in the vagina ( Figure S1). In skeletal muscles, the upregulated GO terms included mitotic nuclear division and mitotic cell cycle, and the downregulated GO terms included immune system process and antigen processing and presentation ( Figure S1).
In addition to the tissue-specific changes in different tissue types, we identified 38 common DE genes in more than five tissues

| Tissue hyperplasia and local immune cell infiltration in KO mice
To confirm the changes in cell proliferation in KO mice, histological staining was performed to further analyse all tissue samples. H&E  Figure 3A); in contrast, other tissues, including those from the skin, rectum and liver, showed no apparent changes.
Next, we performed immunofluorescence staining using the proliferation marker KI67. Compared with the WT mice, the KO mice had higher levels of KI67 expression in the vaginal basal lamina, liver and crypts of the small intestine ( Figure 3B, 3C). This was consistent with the transcriptomic data, in which GO terms related to cell proliferation ( Figure 3D), including keratinization, epidermal cell differentiation and positive regulation of mesenchymal cell proliferation, were upregulated. Immunofluorescence staining for KI67 was also repeated using the remaining tissues; however, no differences were observed between the two groups.

| Systemic immune changes in KO mice detected using CyTOF (mass cytometry)
Based on the upregulation of immune-related GO terms detected using RNA-Seq, we predicted changes in the systemic immunity of KO mice. To evaluate this, we performed CyTOF for characterizing the immune cell populations in peripheral blood samples from WT and KO mice using 32 cell markers. This was performed to assess the effect of Loxl1 inefficiency on systemic immunity. The antibody panel targeted major innate and adaptive immune cell subset markers, including cell surface markers, functional markers and cytokines.
CyTOF data were processed using Cytobank (www.cytob ank.org) and SPADE, which are tools used for the visualization of complex cell composition and proportion in the peripheral blood ( Figure 4A).
Well-recognized cell surface markers were used to identify specific cell subsets in the peripheral blood ( Figure 4B). We first performed gating to obtain CD45+ cells from the peripheral blood samples ( Figure S2A); SPADE analysis revealed that the gated cells were all CD45+( Figure S2B). We then identified major cell subpopu-

| Clinical relevance between LOXL1 mutation and cancer
As LOXL1 deficiency can induce changes in the immune system and tissue hyperproliferation, which are the hallmarks of tumours, 26  were significantly higher in patients with tumours than in the general population ( Figure 5C, 5D), which indicated that a higher LOXL1 mutation frequency may be associated with tumorigenesis.
Next, we compared the overall survival rates between patients with LOXL1-non-mutant and LOXL1-mutant tumours from TCGA.
Among the pan-cancer patients, the overall survival rate in patients with LOXL1-mutant tumours (n = 15) was relatively lower than that in patients with non-mutant tumours (n = 5780) (P =.1438) ( Figure 5E).
The GO analysis revealed that among patients with hepatocellular carcinoma, the cell cycle-related GO terms were significantly upregulated in patients with LOXL1-mutant tumours than in patients with LOXL1-non-mutant tumours, and similar outcomes were observed in GO terms related to mitotic cell cycle, mitotic cell cycle process and cellular response to DNA damage ( Figure 5F). In patients with lung adenocarcinoma, GO terms related to immune activation, including T-cell proliferation, T-cell migration, signal transduction and cytokine production, were significantly upregulated in patients with LOXL1-mutant tumours ( Figure 5G). Therefore, compared to patients with LOXL1-non-mutant tumours, patients with LOXL1-mutant tumours not only had a lower survival rate but also higher levels of transcripts of genes associated with biological processes, such as cell proliferation and immune activation. These results are consistent with the phenotypes identified in Loxl1 KO mice.

| D ISCUSS I ON
Our findings showed that abnormal elastin activity induced by Loxl1 deficiency can induce a series of tissue-specific elastin-related modifications. Using high-throughput sequencing combined with F I G U R E 4 Systemic immune changes in knockout (KO) mice. A, Scheme of the experimental procedure for the characterization of immune cell populations in peripheral blood samples using mass cytometry. Peripheral blood extracted from four mice from each group was pooled into a single sample and stained using a cocktail of metal-tagged antibodies. The samples were analysed using a CyTOF instrument, and the immune cell populations were identified using SPADE and standard dot plots. B, SPADE analysis for the characterization of immune cell populations using canonical markers, including CD3, CD4, CD8, gdTCR, IgM, CD11b, Ly6G, Ly6C, F4/80 and CD49b. In SPADE, the node size correlates with the number of cells, and the coloured gradient corresponds to the arcsinh-transformed expression of the median expression value. Dot plots representing marker labelling on CD45+ cells and the proportion of each cell population in the wild-type and KO groups are shown.  27 The matrix fragment of collagen (the tripeptide N-acetyl Pro-Gly-Pro) plays an important role neutrophil chemoattraction owing to its ability to mimic the chemotactic properties of CXCL8. 28 Other constituents of the ECM, such as hyaluronan, induced the perpetuation of inflammatory responses through the activation of TLR2, TLR4 or both. 29 Fibulin-5 can affect cutaneous inflammation by regulating the NF-κB levels in active inflammation. 30 Therefore, in Loxl1 KO mice, the promotion of the local infiltration of monocytes and macrophages may have been mediated through chemokine-and cytokine-induced immunity.
The immune system plays a crucial role in tissue homeostasis. Our results also revealed a close relationship between Loxl1 mutation and tumour susceptibility, with a higher mutation frequency and lower overall survival rate observed in patients with Loxl1 mutations. The specific cell cycle phase and inflammation in the cellular microenvironment are closely related to cancer progression and outcome 36 as an inflamed microenvironment can disrupt the quiescence of tumour stem cells and initiate tumorigenesis, 37,38 which is consistent with our finding that Loxl1 KO induces local and systemic immune responses as well as multiple tissue hyperplasia.
Previous studies have also shown that increased proliferation and self-renewal play a key role in tumorigenesis. The high rate of DNA replication during proliferation makes the process more error-prone.
Among all cancer-related mutations, 66.1% were attributed to random DNA replication errors (39). Increased mutation rates and genomic instability are inherent characteristics of tumour progression, and an increase in LOXL1 mutations may simply be a red herring rather than a cause of cancer progression. Therefore, further studies are required to confirm whether Loxl1 KO mice have a higher random mutation rate.

CO N FLI C T O F I NTE R E S T
All authors have agreed to the publication of this paper and declare no potential conflicts of interest.

AUTH O R CO NTR I B UTI O N S
XHZ had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data in this paper are deposited in a public database and will be made available upon publication.