Pericyte dysfunction due to Shb gene deficiency increases B16F10 melanoma lung metastasis

Intravasation, vascular dissemination and metastasis of malignant tumor cells require their passage through the vascular wall which is commonly composed of pericytes and endothelial cells. We currently decided to investigate the relative contribution of these cell types to B16F10 melanoma metastasis in mice using an experimental model of host Shb gene (Src homology 2 domain‐containing protein B) inactivation. Conditional inactivation of Shb in endothelial cells using Cdh5‐CreERt2 resulted in decreased tumor growth, reduced vascular leakage, increased hypoxia and no effect on pericyte coverage and lung metastasis. RNAseq of tumor endothelial cells from these mice revealed changes in cellular components such as adherens junctions and focal adhesions by gene ontology analysis that were in line with the observed effects on leakage and junction morphology. Conditional inactivation of Shb in pericytes using Pdgfrb‐CreERt2 resulted in decreased pericyte coverage of small tumor vessels with lumen, increased leakage, aberrant platelet‐derived growth factor receptor B (PDGFRB) signaling and a higher frequency of lung metastasis without concomitant effects on tumor growth or oxygenation. Flow cytometry failed to reveal immune cell alterations that could explain the metastatic phenotype in this genetic model of Shb deficiency. It is concluded that proper pericyte function plays a significant role in suppressing B16F10 lung metastasis.

For tumor cells to reach the vascular circulation and metastasize, they must become invasive, transmigrate through the layer of pericytes/mural cells surrounding endothelial cells (EC) and penetrate the endothelial barrier. 1 After dissemination, the tumor cells must at sites of metastasis again cross the vasculature before they can seed at permissive sites. 1 Although endothelial leakage has received considerable attention as a factor contributing to metastasis, 2-5 pericyte function has also been suggested to play a role in this context. 6,7 Problems in assessing the relative contributions of EC and pericytes lie in the fact that these cell types are reciprocally interdependent [8][9][10] and that platelet-derived growth factor receptorβ (PDGFRB) expressing tumor stromal cells affect the properties of the tumor cells themselves. 11 Melanomas are highly invasive and metastatic. [12][13][14] Previously, their prognosis was poor but has recently improved markedly due to advances in immunotherapy. 15,16 Despite tremendous therapeutic success, a number of cases remain refractory to treatment and thus improved understanding of the mechanisms of melanoma metastasis is warranted.
The Shb gene codes for Src homology-2 domain-containing protein B (SHB) which is an adapter protein downstream of several tyrosine kinase receptors such as vascular endothelial growth factor receptor-2 (VEGFR2) and PDGFRB. 17 SHB has been found to play a role in various aspects of tumor biology. These include tumor angiogenesis [18][19][20] and immune cell responses. 21 Absence of SHB also influences the characteristics of myeloid and lymphoid leukemia. [22][23][24] Of relevance to tumor angiogenesis, SHB is required for VEGF-induced vascular leakage 25,26 and immune cell infiltration. 26 B16F10 melanomas grown in Shb+/− mice exhibit certain intricate and contradictory features. 20 That study demonstrated reduced pericyte coverage of small patent venule-like vessels without detectable effects on tumor angiogenesis. 20 Despite increased leakage, CD8a+ cell infiltration was reduced, 20 an effect probably dependent on cell autonomous effects of Shb deficiency in immune cells. 27 Although bone marrow transplantation experiments 20 suggested that the main contribution to the increase in metastasis related to the host genotype (ie, primarily endothelial cells or pericytes), there could have been an immune cell contribution to the increase in metastasis as well, although that would have been of lesser importance.
We currently set out to elucidate the relative importance of Shb deficiency in endothelial cells (EC) and pericytes by usage of conditionally inactivated Shb flox/flox 27 in EC (Cdh5-CreERt2) 28 and pericytes (Pdgfrb-CreERt2). 29 Our findings suggest that pericyte dysfunction plays a major role in melanoma metastasis.

| Mice
Shb flox/flox mice have previously been described 27 and were bred with Cdh5-CreERt2 mice which were kindly provided by Dr Ralf Adams 28 for Shb gene inactivation in EC. Pdgfrb-CreERt2 29 transgenic mice (https://www.jax.org/strain/029684) obtained from the Jackson Laboratories (Bar Harbor, Maine) were crossed with Shb flox mice to obtain wild-type Pdgfrb-CreERt2 and Shb flox/flox /Pdgfrb-CreERt2 mice for conditional Shb inactivation in mural cells. Cre activity was induced in mice aged 6 to 8 weeks via pretreatment with five daily intraperitoneal injections of 2 mg tamoxifen in peanut oil.
All animal experiments were approved by the local animal ethics committee at the Uppsala County Court (C103115/15).

| Tumor studies
All experiments were performed with mycoplasma-free cells. B16F10

| Cell isolation
Tumors and lungs were excised and digested with collagenase, and a single cell suspension was prepared as described previously. 27 Alternatively, cell suspensions were prepared from spleens, local lymph nodes and blood as previously described. 27 After lysis of red blood cells, the cell suspension was incubated with CD11b+ microbeads (Miltenyi Biotec, see Table S2  microbeads, and further column purified as above for CD31+ cells. To enrich for pericytes, the cell suspension from lungs was incubated with biotin-anti-CD45 followed by anti-biotin microbeads. The flowthrough, devoid of CD45+ cells purified on the magnetic separation columns, was collected and incubated with biotin-anti-CD80 plus antibiotin microbeads to purify CD80+ cells predepleted of CD45+ cells by a second round of selection on a magnetic separation column.

| Immunofluorescence
Resected primary tumors were frozen on dry ice for immunofluorescence staining. Tumor sections (10 μm) were fixed in cold 4% PFA for 10 minutes, permeabilized in 0.2% Triton-X, blocked with 3% BSA and then stained with the following: rat anti-CD31, goat anti-VE-cadherin (vascular endothelial-cadherin, see Table S2 for specifications of reagents), rabbit anti-desmin and goat anti-fibrinogen. Donkey secondary antibodies were used for detection.
Confocal fluorescence images were acquired with a Zeiss LSM780 laser-scanning confocal microscope (Carl Zeiss Microscopy, LLC, Thornwood, New York). The percent area density of blood vessels, the fibrin spread area, hypoxic area relative values by HIF1a staining intensity and pericyte coverage were calculated with Image J by marking relevant areas (whole image field for CD31+ density and HIF1a intensity or for fibrinogen leakage CD31+ enclosed area or area containing surrounding fibrinogen positive staining) after which staining intensity was measured by integrated density.

| Semiquantitative real-time RT-PCR and RNAseq
Total RNA from tumors or magnetic cell sorting (MACS)-isolated cells was isolated using the RNeasy mini-kit (Qiagen, Hilden, Germany). EC (CD31+ cells after predepletion of CD11+ cells) RNA concentration and quality were assessed using the Agilent 2100 Bioanalyzer, and One-step quantitative real-time RT-PCR was performed with QuantiTectTM SYBRGreen RT-PCR-kit (204243, Qiagen, Hilden, Germany) on a LightCyclerTM real-time PCR machine (Lightcycler 2.0; Roche, Mannheim, Germany). Primer sequences are listed in Table S3.
For RNAseq, paired-end sequences were generated using HiSeq2500 high output mode or with Novaseq S1 on libraries made by TruSeq stranded total RNA kit with RiboZero depletion. RNA sequencing was performed at the SNP&SEQ platform, Uppsala University, Sweden. The sequence reads were aligned to the Ensembl mouse gene assembly (GRCm38) using Tophat2 software (version 2.1.1), and FPKM quantification was performed using the Cufflinks tool (version 2.2.1).

| Statistics
Statistical analyses were performed using GraphPad Prism 7.  The data suggest that host Pdgfrb expressing cells play a significant role in melanoma metastasis.

| Tumor vasculature
Neither the EC-specific, nor the pericyte-specific Shb inactivation, exerted an effect on melanoma vascular density ( Figure S2)

| Immune cell alterations
We previously reported immune cell alterations in B16F10 melanomas as a consequence of host Shb gene deletion (Shb +/−) that

| RNAseq of tumor CD31+ RNA
To investigate further the observed decreased vascular leakage in melanomas grown on mice with EC-specific Shb deficiency, EC from B16F10 melanomas grown on wild-type mice or mice with Shb conditionally inactivated in EC (tamoxifen-treated Shb flox/flox /Cdh5-CreERt2) were isolated and subjected to RNAseq (Table S1). Changes in gene expression that were significant by a t test without correction for multi- Anchoring junction P = .0039 21 575 Focal adhesion P = .023 16 411 Cell-substrate adherens junction P = .026 16 416 Cell-substrate junction P = .030 16 421 See Table S3 for complete analysis.
Biological processes that were significantly enriched were, among others, related to the vasculature, circulation, vascular function and vascular development (Table S1). Enriched cellular components were junctions (adherens junctions) and focal adhesions (Table 1) by GO analysis.
These are both key components regulating vascular leakage and thus provide a structural basis for the reduction of tumor leakage shown under these conditions ( Figure 3). Consequently, we investigated the morphology of adherens junctions by staining tumors for VE-cadherin ----F I G U R E 5 Adherens junctions in B16F10 melanomas. Tumors grown on wild-type (wild-type/Cdh5-CreERt2 or wild-type/Pdgfrb-CreERt2) or mice with Shb deficient EC (Shb flox/flox / Cdh5-CreERt2) or pericytes (Shb flox/flox / Pdgfrb-CreERt2) mice were stained for VE-cadherin. The lengths of adherens junctions in vascular structure were measured and normalized to the length of that vascular structure. Values are ratios as means ± SEM for 3 to 4 mice (tumors) in each genotype and * indicates P < .05 when compared to wild-type using a Student's t test. Representative images are shown. The corresponding images with merged CD31 signal can be seen in Figure S5. and observed an increased ratio VE-cadherin positive length per unit vessel length in Shb deficient EC ( Figure 5) which is in line with the functional (leakage) and structural data (RNAseq). Four genes belonging to the GO component "junctions" of which three also belong to "focal adhesions" were confirmed to be increased by qPCR in CD31+ RNA from EC with Shb conditionally deleted ( Figure S4). The increase in Cdh5 mRNA in Shb deficient tumor EC assessed by qPCR was too small to achieve statistical significance (125% of wild-type, P = .38).
RNAseq data analysis supports a role of SHB in EC junction and focal adhesion function/morphology correlating with decreased leakage.

| Tumor oxygenation
Due to dysfunctional pericytes as a consequence of Shb deficiency, tumor oxygenation could decrease, thus causing a metastatic phenotype. 6 Consequently, we stained tumor sections of the genotypes ( Figure 6) for hypoxia-inducible factor 1 alpha (HIF1a