Background & Summary

OSM plays an important role in development, malignancy, and homeostasis of various tissue types1,2,3,4,5. OSM is primarily produced by mature activated hematopoietic cells such as monocytes, T lymphocytes, neutrophils, macrophages, and dendritic cells4,6,7. OSM was first isolated based on its ability to inhibit the proliferation of melanoma tumor cells, while having no significant effect on normal human fibroblasts6. Despite this initial identification of OSM as a tumor suppressor, a pro-inflammatory role for OSM has subsequently been reported in several solid tumors, myeloma, and chronic lymphocytic leukemia1,3,8.

Human OSM interacts with two distinct receptor complexes. The type I receptor complex contains leukemia inhibitory factor receptor alpha (LIFR) and interleukin 6 signal transducer (IL6ST; GP130), while the type II receptor complex contains oncostatin M receptor (OSMRβ) and IL6ST4,9,10,11. While recent studies have shown that OSM has a lower affinity for the type I receptor complex than the type II receptor complex12,13, much remains unknown about how OSM interacts with these complexes in different contexts. While human OSM is known to signal through both the type I and type II receptor complexes, mouse OSM is reported to signal only through the type II receptor complex14,15,16. With respect to hematopoiesis, increased transcript levels of OSMRβ (Osmr) are a significant alteration in mouse HSC aging17,18, prompting us to query the molecular consequences of OSM signaling on HSCs and the extent to which this is dependent on OSMRβ. Thus, we obtained a B6;129-Osmrtm1.1Nat/J (referred to as Osmrfl/fl) mouse model and crossed these mice to an interferon-inducible Mx1-Cre allele19, which is commonly used in the study of adult hematopoiesis and also acts with variable efficiency in other tissues. From these mice and their Mx1-Cre control counterparts, we prospectively isolated HSCs by flow cytometry, stimulated HSCs in vitro with recombinant murine OSM (500 ng/ml) or vehicle for 1 hour and assessed gene expression by RNA-seq (Fig. 1). This RNA-seq data may be used to investigate OSMRβ -dependent and -independent OSM signaling as well as the transcriptional effects of an IL-6 family cytokine on mouse HSCs to further define its anti-inflammatory versus pro-inflammatory properties.

Fig. 1
figure 1

Overview of the study design and experimental workflow. RNA-seq data was generated from hematopoietic stem cells (HSCs) isolated from control Mx1-Cre and experimental OsmrΔ/Δ Mx1-Cre mice that had been treated ex vivo for 1 hour with recombinant murine OSM (500 ng/ml) or vehicle control20. An independent cohort of mice was used for technical validation of the genotype and recombination of OsmrΔ/Δ Mx1-Cre as well as validation of reduced OSMRβ expression by Western blotting. Created with BioRender.com.

Full size image

Methods

Mice

C57BL/6 J (JAX:000664) mice were obtained from The Jackson Laboratory. B6;129-Osmrtm1.1Nat/J (JAX: 011081) were crossed to B6.Cg-Tg(Mx1-cre)1Cgn/J mice (JAX:003556). To induce Mx1-Cre, mice were intraperitoneally injected once every other day for five total injections with 15 mg/kg high molecular weight polyinosinic-polycytidylic acid (polyI:C) (InvivoGen). Mice used for experiments were young female adults (2–4 months of age) and were used >10 weeks after polyI:C administration. The Jackson Laboratory’s Institutional Animal Care and Use Committee approved all experiments.

Genotyping and recombination PCR

Genomic DNA was extracted from lysed peripheral blood samples using the DNeasy Blood & Tissue Kit (Qiagen) for genotyping. Genotyping primers were used as suggested on the B6;129-Osmrtm1.1Nat/J (JAX: 011081) mice webpage (protocol ID 29086 and 23692). Images were taken using the InGenuis LHR2 Gel imaging system (Syngene). Recombination of the Osmrfl locus was evaluated by PCR using the following primers: Forward: 5’-GGAAATACCTTGGCAGTGGTG and Reverse: 5’-GCTACCAAACCTCGGTAATCC.

Western blotting

Total protein was extracted from snap frozen tissues with a Tissue Lyser II (Qiagen) using 5 mm Bead Beater balls (Qiagen) after the addition of 50 mM HEPES pH 7.5 (Gibco) and 6 M urea (Sigma Aldrich) at 1.5 mL per 200 mg tissue. Insoluble material was spun out at 21,000 × g for 20 m. Total protein concentration of each sample was determined by microBCA (ThermoFisher). SDS-PAGE gel samples were prepared. Two 4–15% Mini-PROTEAN TGX Precast, 1.0 mm, 15-well, SDS-PAGE Gel (BioRad) were loaded with 25 ug total protein per well. The gels were run at 150 V for 1 h at room temperature in Tris·Glycine·SDS buffer (BioRad) then transferred overnight at 4 °C at 40 mAmp to 0.22 mm PVDF (BioRad) in Tris·Glycine buffer (BioRad) with 10% methanol. The blots were placed in Block (2% BSA + 4% Dry Milk) at room temperature for 1 h. The blots were then placed in primary antibody Polyclonal Goat IgG anti-mouse OSMRβ (R&D systems AF662) diluted in fresh block at 1:2500 and incubated on an orbital shaker at 4 °C overnight. Then the blots were washed with TBS with 0.03% Tween 20 (TBS-T) for 15 min and three more washes for 5 minutes each. Then the blots were placed in Peroxidase AffiniPure Bovine Anti-Goat IgG (H + L) (Jackson Immuno Research) diluted 1:5,000 in block. The blots rotated for 1 h at room temperature and then were washed with TBS-T for 15 min and three more washes for 5 minutes each. The blot was then developed with (ThermoFisher) SuperSignal West Pico Plus reagents (ThermoFisher) and visualized with a Syngene G-Box. The blots were then stripped with Restore™ PLUS Western Blot Stripping Buffer (ThermoFisher) at room temperature for 5 min. The blots were re-blocked and then incubated with anti-β-actin (Cell Signaling Technology clone D6A8) at 1:5000 for 1 h at room temperature. Then the blots were washed with TBS-T for 15 min and three more washes for 5 minutes each. Then the blots were placed in Goat Anti-Rabbit IgG (H + L)-HRP (Bio-Rad) diluted 1:5,000 in block. The blots rotated for 1 h at room temperature and then were washed with TBS-T for 15 min and three more washes for 5 minutes each. The blot was then developed with SuperSignal West Femto Maximum Sensitivity reagents (ThermoFisher) and visualized with a Syngene G-Box. Images were quantified by ImageJ by relative intensity, adjusted for background and normalized to β-Actin expression per sample. Significance was calculated using paired t test with correction for multiple comparisons using the Holm-Sidak method. Calculation of significance and graphing was performed using Prism 10 (GraphPad).

RNA sequencing

Bone marrow (BM) cells were isolated from OsmrΔ/Δ Mx1-Cre and Mx1-Cre from pooled and crushed femurs, tibiae, iliac crests, sternums, forepaws, and spinal columns of individual mice. BM mononuclear cells (MNCs) were isolated by 1X Red Blood Cell lysis Buffer (eBioscience) and stained with a panel of fluorochrome-conjugated antibodies against c-Kit (BD Biosciences, BioLegend clone 2B8), Sca-1 (BioLegend clone D7), CD150 (BioLegend clone TC15-12F12.2), CD48 (BioLegend clone HM48-1), FLT3 (BioLegend clone A2F10), (Lin) marker mix (B220 (BD Biosciences, BioLegend clone RA3-6B2), and 4′,6-diamidino-2-phenylindole (DAPI). 2,000 hematopoietic stem cells (HSCs) were prospectively isolated based on the cell surface marker combination: Lin- Sca-1+ c-Kit+ Flt3- CD150+ CD48- on a FACSAria with >95% purity (Fig. 2). HSCs were directly sorted into StemSpan SFEM II media with SCF (100 ng/ml, BioLegend), TPO (50 ng/ml, Peprotech), with or without OSM (500 ng/ml, BioLegend) and incubated at 37 °C for 60 min. Cells were then washed, pelleted and flash frozen. Total RNA was isolated from flash-frozen pellets using the RNeasy Micro Kit (Qiagen) including the optional DNase digest step. RNA concentration and quality were assessed using the RNA 6000 Pico Assay (Agilent Technologies). Libraries were constructed using the SMARTer Stranded Total RNA-Seq Kit v2-Pico (Takara) according to the manufacturer’s protocol. Library concentration and quality were assessed using the D5000 ScreenTape (Agilent Technologies) and Qubit dsDNA HS Assay (ThermoFisher). Libraries were subject to 75 bp paired-end sequencing on an Illumina NextSeq 500 using the High Output Reagent Kit v2.5 or 150 bp paired-end on an Illumina NovaSeq 6000 using the S4 Reagent Kit v1.5. Trimmed alignment files were processed using RSEM (v1.3.3 or v1.3.1). Alignment to the mm10 reference genome was completed using Bowtie 2 (v2.4.1 or v2.4.5). Expected read counts per gene produced by RSEM were rounded to integer values. Filtered total read counts had an average of 151 million counts (±32 million standard deviation), Q30 rate of 93% (±1% standard deviation), 52% GC rate (±2% standard deviation), and 28 million (±9 million standard deviation) read pairs mapped to genes (Table S1, see supplementary xlsx file). Counts were filtered to include only genes that had at least two samples within a sample group having a counts per million reads >1 in R (v4.1.3), and edgeR (v3.36.0) was used for differential expression analysis. A negative binomial generalized log-linear model was fit to the read counts for each gene. The dispersion trend was estimated by Cox-Reid approximate profile likelihood followed by empirical Bayes estimate of the negative binomial dispersion parameter for each tag, with expression levels specified by a log-linear model. Likelihood ratio tests for coefficient contrasts in the linear model were evaluated producing a p value per contrast. The Benjamini and Hochberg’s algorithm was used to control the false discovery rate. Features with fold change (FC) >2 or <−2 and p < 0.05 were declared significantly differentially expressed. Predicted genes and pseudogenes were removed from the list of significantly differentially expressed genes before downstream comparisons were performed.

Fig. 2
figure 2

Cell sorting plots and purity for control Mx1-Cre and OsmrΔ/Δ Mx1-Cre HSCs. (A) Representative flow cytometric gating used for isolation of HSCs as viable Lin- Sca-1+ c-Kit+Flt3- CD150+ CD48- cells. (B) Representative purity of isolated Lin- Sca-1+ c-Kit+ HSPCs after cell sorting, which was >95%. (C) Representative purity of isolated Lin- Sca-1+ c-Kit + Flt3- CD150+ CD48- HSCs after cell sorting, which was >99%.

Full size image

Data Records

The RNA-seq dataset is available at the Gene Expression Omnibus (GEO) [GSE24454420], with this section being the primary source of information on the availability and content of the data being described. This dataset includes a total of 18 RNA-seq samples. The first group of samples are control Mx1-Cre HSCs treated for 1 hour with vehicle (n = 6, filenames: GSM7818933 Mx1-Cre Veh 1, GSM7818934 Mx1-Cre Veh 2, GSM7818935 Mx1-Cre Veh 3, GSM7818936 Mx1-Cre Veh 4, GSM7818937 Mx1-Cre Veh 5, GSM7818938 Mx1-Cre Veh 6)20. The second group of samples are control Mx1-Cre HSCs treated for 1 hour with 500 ng/ml OSM (n = 6, filenames: GSM7818939 Mx1-Cre OSM 1, GSM7818940 Mx1-Cre OSM 2, GSM7818941 Mx1-Cre OSM 3, GSM7818942 Mx1-Cre OSM 4, GSM7818943 Mx1-Cre OSM 5, GSM7818944 Mx1-Cre OSM 6)20. The third group of samples are OsmrΔ/Δ Mx1-Cre HSCs treated for 1 hour with vehicle (n = 3, filenames: GSM7818927 Osmrfl/fl Mx1-Cre Veh 1, GSM7818928 Osmrfl/fl Mx1-Cre Veh 2, GSM7818929 Osmrfl/fl Mx1-Cre Veh 3)20. The fourth group of samples are OsmrΔ/Δ Mx1-Cre HSCs treated for 1 hour with 500 ng/ml OSM (n = 3, filenames: GSM7818930 Osmrfl/fl Mx1-Cre OSM 1, GSM7818931 Osmrfl/fl Mx1-Cre OSM 2, GSM7818932 Osmrfl/fl Mx1-Cre OSM 3)20.

We utilized this dataset to explore how control Mx1-Cre versus OsmrΔ/Δ Mx1-Cre HSCs respond to stimulation with recombinant murine OSM. First, we performed principal component analysis to cluster control Mx1-Cre and OsmrΔ/Δ Mx1-Cre HSCs treated with or without OSM20 (Fig. 3A). We observed that OSM-treated versus vehicle-treated control HSCs clearly separated on principal component 1 (PC1) whereas OSM-treated versus vehicle treated OsmrΔ/Δ HSCs did not separate on PC1 and were partially separated on PC2, indicating greater transcriptional changes in control HSCs is response to OSM compared to OsmrΔ/Δ HSCs in response to OSM. Using the cutoffs for significance defined above, control HSCs had 171 genes increased in expression following OSM stimulation compared to vehicle treatment and 151 genes decreased in expression following OSM stimulation (Fig. 3B, Table S2, Table S3, see supplementary xlsx files). These differentially expressed genes were observed across biological replicate samples (Fig. 3C). Focusing on the 171 genes induced by OSM in control HSCs, we examined their expression in OsmrΔ/Δ HSCs (Table S4). This revealed several patterns: 151 of these genes (88.3%) were no longer induced by OSM in OsmrΔ/Δ HSCs, 7 genes (4.1%) were induced by OSM in both control and OsmrΔ/Δ HSCs, and 5 genes (2.9%) were induced by OSM in control HSCs but repressed by OSM in OsmrΔ/Δ HSCs (Fig. 3D). We then analyzed the 151 genes induced by OSM stimulation in control HSCs but no longer induced in OsmrΔ/Δ HSCs using the molecular signatures database (MSigDB)21 focused on gene ontology (GO) terms. This analysis revealed enrichment of GO terms including cell motility, protein containing complex localization, cell adhesion, signaling receptor binding and positive regulation of gene expression (Fig. 3E).

Fig. 3
figure 3

Differential gene expression analysis of recombinant OSM-stimulated genes in control Mx1-Cre and OsmrΔ/Δ Mx1-Cre HSCs. (A) Principal component analysis of all biological replicate samples of control Mx1-Cre HSCs treated with OSM (n = 6) or without OSM (n = 6) and OsmrΔ/Δ Mx1-Cre HSCs treated with OSM (n = 3) or without OSM (n = 3)20. (B) Volcano plot showing differential gene expression of OSM-treated vs. vehicle-treated control Mx1-Cre HSCs. Numbers of significantly differentially expressed genes (log2(FC) >2 or <−2, p < 0.05) are indicated. See also Supplementary Table 2. (C) Heatmap of gene expression across all samples of the 322 differentially expressed genes identified in OSM-treated vs. vehicle-treated control Mx1-Cre HSCs. (D) Proportions of the 171 OSM-induced genes in Mx1-Cre HSCs assessed in OSM-treated vs. vehicle-treated OsmrΔ/Δ Mx1-Cre HSCs. Cutoff for significantly induced or repressed genes utilized p < 0.05 criteria. See also Supplementary Table 3. (E) Gene ontology (GO) term enrichment of 151 OSM-induced genes in control Mx1-Cre HSCs that were not induced by OSM in OsmrΔ/Δ Mx1-Cre HSCs.

Full size image

Technical Validation

Verification of genotype and osmr tm1.1Nat allele recombination

After crossing B6;129-Osmrtm1.1Nat/J to Mx1-Cre to generate Osmrfl/fl Mx1-Cre and control Mx1-Cre animals, we genotyped three control and three experimental animals alongside positive (parental Osmrfl/fl and Mx1-Cre mice) and negative (wild-type C57BL/6 J) controls. At the Osmr locus, genomic DNA from Osmrfl/fl Mx1-Cre mice contained a loxP site (230 bp) and genomic DNA from control Mx1-Cre mice did not contain a loxP site (170 bp), recognized by primers flanking exon 2 (Fig. 4A). Genomic DNA from all Osmrfl/fl Mx1-Cre and control Mx1-Cre mice were also positive for the transgenic cre allele (100 bp) (Fig. 4B). Following injection with poly(I:C) to induce Mx1-Cre-mediated recombination of the Osmrfl locus, we isolated DNA from bone marrow, liver, and kidney tissues. A recombination PCR assay was designed to assess Osmrfl recombination efficiency. The primers generate a product spanning the loxP-flanked exon 2 and bind to both the wild-type and modified Osmr loci generating product sizes of ~860 bp for the floxed (fl) allele, 738 bp for the wild-type (wt) allele and ~300 bp for the recombined (Δ) allele. This PCR assay detected a ~300 bp recombined allele in all tissue samples from OsmrΔ/Δ mice (Fig. 4C), demonstrating that recombination of the Osmr locus occurred efficiently following poly(I:C) injection in the bone marrow, liver, and kidney.

Fig. 4
figure 4

Genotyping and recombination of the Osmrfl locus in tissues from OsmrΔ/Δ Mx1-Cre mice. Agarose gel electrophoresis (2% agarose) of PCR-amplified products. In all gels a 100 bp ladder was used for size estimation located in lane 0. (A) Genotyping PCR for Osmr <tm1.1Nat> from ear punch genomic DNA. Lanes 1–3 are from Mx1-Cre mice, lanes 4–6 are from Osmrfl/fl Mx1-Cre mice, lane 7 is from a wild-type C57BL/6 J mouse and lane 8 is from a positive control Osmrfl/fl confirmed by The Jackson Laboratory’s in-house genotyping service. (B) Genotyping PCR for Cre recombinase from ear punch genomic DNA. Lane 1 is from a positive control, lane 2 is from a wild-type C57BL/6 J mouse, lanes 3–5 are from Mx1-Cre mice and lanes 6–8 are from Osmrfl/fl Mx1-Cre mice. (C) Recombination PCR for the Osmr <tm1.1Nat> allele from indicated tissues. From left, lanes 1–2 are from blood samples from wild-type C57BL/6 J mice. In the remaining images, samples are ordered as lanes 1–3 from Mx1-Cre mice and lanes 4–6 from OsmrΔ/Δ Mx1-Cre mice.

Full size image

Confirmation of OSMRβ knockout

We performed western blotting with a published murine OSMRβ antibody12,14,22,23 to evaluate protein expression in the liver, kidney, and bone marrow isolated from OsmrΔ/Δ and control mice. OSMRβ in the liver was undetectable in OsmrΔ/Δ mice (Fig. 5A) and OSMRβ expression in the kidney and bone marrow were reduced to very low levels in OsmrΔ/Δ mice (Fig. 5B,C). Quantitation of bands by relative intensity using ImageJ determined that OSMRβ was significantly reduced in OsmrΔ/Δ compared to control mouse liver, kidney, and bone marrow, respectively (Fig. 5D). Due to the variable efficiency in Mx1-Cre-mediated deletion across tissues, OSMRβ expression may be retained on certain cells, accounting for differences between the data presented in Figs. 4 and 5.

Fig. 5
figure 5

Assessment of OSMRβ in tissues from OsmrΔ/Δ Mx1-Cre mice. (A–C) Western blot analysis of cell lysates from (A) liver, (B) kidney, and (C) bone marrow from Mx1-Cre and OsmrΔ/Δ Mx1-Cre mice. Blots were probed with antibodies against OSMRβ (~150 kDa) and β-ACTIN (45 kDa) (n = 3 biological replicates per genotype). (D) Relative protein expression assessed by Western blot imaging using ImageJ. OSMRβ band densities were normalized to β-ACTIN. Dots represent individual mice. *p < 0.05. Significance was calculated using paired t test with p values adjusted for multiple comparisons using the Holm-Sidak method.

Full size image