We previously demonstrated that transplantation of 4T1 breast cancer cells to BALB/c female mice strongly increased the expression of Saa1-2 mRNAs in the liver (14). We re-analyzed the RNA-seq datasets we recently published (20) to confirm that Saa1-2 mRNAs were induced in the livers of 4T1 breast cancer-bearing mice ( Figure 1A ). We further validated this observation using reverse transcription quantitative PCR (RT-qPCR) detecting both Saa1 and Saa2 whose nucleotide sequences are 95% identical ( Figure 1B and Figure S1A and Table S1 ). We also found that 4T1 breast cancers elevated hepatic SAA1-2 at the protein level ( Figure 1C ). Together, we concluded that transplantation of 4T1 breast cancers enhanced acute phase response in the liver, which is in line with other cancer models (13).

To address whether the increased expression of SAA1-2 affects inflammation, we characterized our liver transcriptome data using gene set enrichment analysis (GSEA) ( Figure 1D ) (21, 22). Our analysis revealed that 4T1 breast cancer triggered various inflammatory responses in the livers as exemplified by the activation of the interleukin-6 (IL-6) signaling (23). This was in line with the previous observations that Saa1-2 genes are under the control of the IL-6 signaling (13) and that solid cancers instigate the IL-6 signaling in the liver (3, 23). We then used ImmuCellAI-mouse to deduce the infiltration of various immune cells into the liver (24, 25). Using this method, we found that neutrophils migrated into the liver upon cancer transplantation ( Figure 1E ), as previously reported (13, 14). Furthermore, we noted the close correlations at the mRNA level between Saa1-2 and other neutrophil-related genes such as Lcn2 ( Figure 1F and Figure S1B ). Together with the known biochemical roles of SAA1-2 (9), these results led to a hypothesis that SAA1-2 proteins play some roles in immune cell activation in the presence of 4T1 breast cancers.

To uncover the roles of SAA1-2 in inflammation in vivo in this model, we generated mice completely lacking Saa1 and Saa2. These two genes are located closely in the murine genome ( Figure 2A ), having redundant sequences and molecular functions ( Figure S1A ) (9). We thus decided to delete the entire region encoding Saa1 and Saa2 genes by designing gRNAs on the right and left sides of this genomic locus ( Figure 2A ) (26). As a result, we succeeded in deleting both Saa1 and Saa2 as determined by genomic PCR ( Figures 2A, B ). Transplantation of 4T1 breast cancer cells to Saa1-2 knockout mice no longer increased the expression of Saa1-2 mRNAs ( Figure 2C ) and proteins ( Figure 2D ) in the liver. Western blot experiments demonstrated that two bands detected in lysates prepared from the livers of 4T1-bearing mice disappeared in Saa1-2 knockout mice ( Figure 1C and Figure 2D ). Adding recombinant SAA1 protein that lacks the signal peptide as a control, we reasoned that the lower band corresponds to the cleaved SAA proteins ( Figure S2 ). Taken together, we established mice where we completely canceled cancer-dependent increase in SAA1-2 proteins in the liver, allowing us to investigate the in vivo significance of these proteins in cancer-induced liver inflammation.

To thoroughly analyze the effects of Saa1-2 knockout on the liver transcriptome, we performed RNA-seq analyses against the livers of WT and Saa1-2 knockout mice ( Figure 3A and Table S2 ). We found that 4T1 breast cancer transplantation similarly affected liver transcriptome regardless of the presence or absence of Saa1-2 genes, as evidenced by the volcano plots shown in Figure 3A and Figure S3A, B . GSEA demonstrated that 4T1 breast cancers could still activate the IL-6 signaling and inflammatory response in the absence of Saa1-2 genes ( Figure 3B ), suggesting a subtle or negligible contribution of Saa1-2 to the liver transcriptome in our experimental settings. In addition, we wanted to confirm these observations using different cohorts of 4T1 transplantation experiments. For this purpose, we quantified the mRNA expression of various immune cell marker genes in the liver, finding that Saa1-2 KO did not have a significant impact on cancer-dependent up-regulation of representative immune-related genes in the liver ( Figure S3C ). These results suggested that 4T1 breast cancer cells do not require Saa1-2 to increase the expression of immune-related genes in the liver in our experimental settings.

A previous study reported the significant contribution of SAA proteins in recruiting innate immune cells including neutrophils to the liver in the pancreatic cancer-bearing condition (13). This prompted us to further investigate the roles of SAA1-2 in immune cell recruitment to the liver in the 4T1 breast cancer model. To this end, we estimated the proportions of several immune cell types in the livers of cancer-bearing mice, comparing them between WT and Saa1-2 KO.

ImmuCellAI-mouse analyses confirmed that transplantation of 4T1 breast cancer cells increased the proportion of neutrophils within the liver ( Figure 4A and Table S3 ). Of note, the proportions of neutrophils in the liver were comparable between WT and Saa1-2 KO in the cancer-bearing condition, implying negligible roles of SAA1-2 in recruiting neutrophils in this model. We also investigated the proportions of neutrophils, monocytes, and macrophages using flow cytometry. We collected immune cells from the livers of sham and cancer-bearing animals, quantifying those immune cells using a set of specific antibodies (see figure legends and methods). As shown in Figure 4B , our data revealed that the proportions of neutrophils, monocytes, and macrophages were unaffected by Saa1-2 KO. Collectively, although 4T1 breast cancer transplantation strongly induced SAA1-2 in the liver, these acute phase proteins do not appear to be essential for cancer-dependent immune cell recruitments into the liver in the 4T1 model ( Figures 1 - 4 ).

Innate immune cells such as neutrophils are born and matured in the bone marrow (27). It is also known that cancers affect immune cell development (4). Given these, it was likely that the altered immune cell status in the liver was owing to changes in the bone marrow. We thus wanted to extend our experiments on the roles of Saa1-2 in immune cell activation in the bone marrow.

We performed RNA-seq experiments against immune cells collected from the bone marrow from WT and Saa1-2 KO mice from which we obtained the liver transcriptome data ( Figure 3 and Table S2 ). We found that 4T1 transplantation affected gene expression in the bone marrow, resulting in many differentially expressed genes ( Figure 5A , Figure S5A , and Table S4 ). According to GSEA, in WT, 4T1 transplantation activated the IL-6 signaling and inflammatory response ( Figure 5B ), which is in line with our liver data ( Figure 3B ). These results implied that inflammatory response observed in the liver is correlated with altered immune cell dynamics in the bone marrow. Moreover, ImmuCellAI-mouse analysis demonstrated that 4T1 transplantation altered the proportions of various cell types including neutrophils in the bone marrow, supporting a cancer-induced reprogramming of the host immune system ( Figure 5C and Table S3 ) (4). Most importantly, none of these immune cell phenotypes in the bone marrow was strongly buffered by the deletion of Saa1-2. We validated these data using qPCR with the different cohorts of experiments ( Figure S5B ). These results together provided evidence that the effects of Saa1-2 KO on the bone marrow transcriptome were negligible or minor if any, suggesting dispensable roles of SAA1-2 in 4T1-induced reprogramming of the host immune system. Thus, our results suggested that SAA1-2 proteins are not essential for liver inflammation observed in the 4T1 breast cancer model.

All animal experiment protocols were approved by the Animal Care and Use committee of Kyoto University. Mice were housed as described previously (20) in a 12-hour light/dark paradigm with food (CE-2, CLEA Japan, Inc., Tokyo, Japan) and water available ad libitum. Mice were randomly assigned to different experimental groups without any specific criterion. No blinding was performed. WT BALB/c mice were purchased from Japan SLC Inc. (Hamamatsu, Japan).

BALB/c Saa1-2 KO mice were generated as described previously (20, 26). In vitro fertilized eggs stored were thawed and electroporated using CUY-EDIT II (BEX, Tokyo, Japan) (amplitude 20V, duration 3 msec., interval 97 msec. for twice) with two independent crRNAs (25 ng/µL: FASMAC, Kanagawa, Japan), tracrRNA (100 ng/µL: FASMAC), and purified recombinant Cas9 proteins (250 ng/µL: Thermo Fisher Scientific, MA, USA). The crRNA sequences are as follows:

Eggs at the 2-cell stage were transplanted into the oviduct of pseudopregnant mice. F0 mice were crossed with WT and F1 mice were obtained for generating KO mice (≥ F2).

Genomic PCR to genotype mice was performed against DNAs prepared from the mouse tails. The tails were incubated with 90 µL of 50 mM NaOH (nacalai tesque, Kyoto, Japan) for more than 10 min at 95°C. 10 µL of 1 M Tris-HCl pH 8.0 (nacalai tesque) was then added to the reaction, followed by centrifugation at 10,000 × g for more than 10 min. The resulting supernatant was subjected to genomic PCR. Genomic PCR for genotyping was performed using KOD FX-Neo (TOYOBO, Osaka, Japan). The primers used in this experiment are shown in Table S1 .

The 4T1 mouse breast cancer cell line (20) was cultured and maintained in RPMI1640 (nacalai tesque) in a 5% CO2 tissue culture incubator at 37°C as described previously (20). The media (RPMI1640) was supplemented with 10% fetal bovine serum (nacalai tesque) and 1% penicillin/streptomycin (nacalai tesque). The thawed cells were passaged once and then were transplanted to mice. 2.5×10⁶ 4T1 cells resuspended in 100 µL of RPMI1640 containing neither FBS nor penicillin/streptomycin were inoculated subcutaneously into the right flank of an 8–9-week-old BALB/c female mouse. In the sham-treated group, mice were given RPMI1640 supplemented with 10% FBS. Mice were sacrificed on day 14 post-transplantation and the liver and bone marrows were collected.

RNA isolation, cDNA synthesis, and RT-qPCR experiments were performed as described previously (20). Mouse livers were crushed in liquid nitrogen and homogenized with Trizol reagent (Thermo Fisher Scientific). Total RNAs were extracted from the homogenized supernatant using RNeasy Mini Kit (Qiagen, Venlo, Netherlands) according to the manufacturer’s instructions.

The bone marrows were collected essentially as described previously (31, 32). Briefly, the mouse right femurs from sham and 4T1-bearing mice were collected, cleaned of muscle tissue, and then were polished with gauze. The femurs were placed with the knee end down in a perforated 0.6-mL tube to which 80 μL of RNAlater (Qiagen) was added and inserted in a 1.5 mL centrifuge tube, followed by centrifugation at 5,700 × g for 30 sec. The resulting pellets were suspended in 1 mL of Trizol reagent (Thermo Fisher Scientific) and proceeded with RNA extraction as described above.

Total RNAs (100 or 500 ng) were reverse-transcribed using Transcriptor First Strand cDNA synthesis kit (Roche, Basal, Switzerland) in a 10 μL reaction, which was then diluted 10-or 50-fold, respectively. qPCR experiments were performed using the StepOnePlus qPCR system (Applied Biosystems, CA, USA) and SYBR Green Master Mix (Roche). We used 2 μL of the obtained cDNA in a 10 μL qRT-PCR reaction. Gapdh was used as an internal control for the liver samples. 18S rRNA was used as an internal control for the bone marrow samples.

Crushed liver powders were lysed with lysis buffer (10 mM Tris-HCl pH8.0, 100 mM KCl, 2.5 mM MgCl₂, 0.5% Triton Χ-100, cOmplete (Roche, Basel, Switzerland). The protein concentration was determined using the BCA protein assay kit (TaKaRa, Shiga, Japan) according to the manufacturer’s instructions. Twenty μg of the extracted protein were electrophoresed on a 15% sodium dodecyl sulfate (SDS) polyacrylamide gel for 1 hour at 150V and transferred to a polyvinylidene fluoride (PVDF) membrane (Millipore, MA, USA) for 1 hour at 72V. The membrane was incubated with 5% skim milk in Tris-buffered saline, 0.1% Tween20 (TBST) overnight at 4°C, and then was incubated with mouse serum amyloid A1/A2 antibody (1:1000 in Can Get Signal (TOYOBO, Osaka, Japan): AF2948, R&D Systems, MN, USA) for 2 hours at room temperature. The membrane was washed with TBST three times. Then, the membrane was incubated with goat IgG horseradish peroxidase-conjugated antibody (1:5000 in Can Get Signal (TOYOBO): HAF017, R&D systems) for 1 hour at room temperature. Following TBST-wash steps, signals were visualized using ECL Prime Western Blotting Detection reagent (Cytiva, Tokyo, Japan) and images were taken using Amersham ImageQuant800. The same protein samples were loaded onto a 15% SDS polyacrylamide gel and then stained with SYPRO Ruby Protein Gel Stain (Thermo Fisher Scientific) to confirm equal sample loading in each lane.

Total RNAs were extracted as described above with RNase-Free DNase Set (Qiagen). RNA-seq libraries were generated using the NEBNext Globin&rRNA depletion kit and the NEBNext UltraII Directional RNA Library prep kit according to the manufacturer’s instructions (New England Biolabs, MA, USA). Sequencing experiments were performed using NextSeq 500 (Illumina; High Output Kit v2.5, 75 cycles). The obtained reads were filtered using fastp (version 0.20.1) (33) to remove low-quality sequences (< Q30), mapped to the mouse genome (version mm10) using Hisat2 (version 2.2.1) (34), and processed using Samtools (version 1.10) (35) and featureCounts (version 2.0.1) (36). Read counts were normalized with the reads per million per kilobase (RPKM) method. The generated gene expression matrix with RPKM scores is listed in Table S2 and Table S4 .

The generated gene expression matrix with RPKM scores was used to perform gene set enrichment analysis (GSEA) to interpret transcriptional profiles using GSEA 4.2.3 software (22, 37). Human hallmark collections were obtained from the Molecular Signatures Database (MSigDB, v7.5.1). A total of 1000 permutations were set to determine the significance of the enrichment for the gene sets, and the “Signal to noise” or “log₂ ratio of classes” metric was conducted to rank the genes depending on sample size in each group according to the instruction. Enrichment score (ES), normalized enrichment score (NES), false discovery rate (FDR) and signatures were obtained. The enriched gene sets were defined as significant with FDR < 0.25. For differentially expressed gene (DEG) analyses, a read count matrix and a condition label vector were taken as input using the R package, edgeR (version 3.40.0) (38). First, we filtered out genes with very low expressions using CPM values rather than counts since they account for differences in sequencing depth between samples. Genes with CPM > 1 in at least 2 or more samples were taken into account. The trimmed mean of M values (TMM) method was applied to normalize the counts of retained genes among the different samples. The exact test function was performed to detect significantly expressed genes. The differentially expressed genes (DEGs) were defined as genes with |log2(fold change)| >1 and p value < 0.05. The volcano plots were depicted using ggplot2 to visualize DEGs (https://ggplot2.tidyverse.org/index.html). ImmuCellAI-mouse (24, 25), a bulk RNA-seq data deconvolution approach, was applied to estimate the abundance of 36 immune cell types using the default settings ( Table S3 ).

The livers of sham and 4T1-bearing mice were harvested on day 14 after transplantation. The obtained liver tissues were homogenized in 8 mL of RPMI1640 media containing 1.8 mg/mL Collagenase IV (WOR-CLS4-1, Worthington Biochemical Corporation, NJ, USA) and 112.3 µg/mL DNase I (11284932001, Roche). 8 mL of the suspension was filtered using a cell strainer (70 µm mesh) and then mixed with 4 mL of 90% Percoll solution (Sigma-Aldrich, MO, USA). The suspension was then centrifuged at 700 × g for 20 min. The red blood cells in the pellets were lysed with 1×Lysing buffer (Lysing Buffer 10× Concentrate (BD Biosciences) diluted by H₂O (nacalai tesque)). The samples were washed with RPMI1640 media supplemented with 2% FBS. The obtained samples were stained for 15 min on ice in FACS buffer (2% FBS, 0.05% NaN_3, and 1×PBS) mixed with TruStain FcX™ (anti-mouse CD16/32) Antibody (1:200, Clone: 93, BioLegend, CA, USA) and eBioscience™ Fixable Viability Dye eFluor™ 780 (1:1000, Invitrogen, MA, USA). The samples were washed with 100 μL of FACS buffer. The washed samples were stained for 20 min on ice with Brilliant Violet 510™ anti-mouse CD45 Antibody (1:200, Clone: 30-F11, BioLegend), PE/Cyanine7 anti-mouse/human CD11b Antibody (1:200, Clone: M1/70, BioLegend), PE anti-mouse Ly-6C Antibody (1:200, Clone: HK1.4, BioLegend), and FITC anti-mouse Ly-6G Antibody (1:200, Clone: 1A8, BioLegend), and APC anti-mouse F4/80 Antibody (1:50, Clone: BM8, BioLegend) in FACS buffer. Following a wash step using FACS buffer, the stained samples were resuspended with FACS buffer and then filtered using a cell strainer (35 μm mesh) set in a 5 mL tube (Falcon). The resulting samples were analyzed using FACS Canto II (BD Bioscience, NJ, USA) and analyzed using FlowJo software (v10.801) (BD Biosciences).

GraphPad Prism Software was used to analyze data. Data were displayed as mean ± SEM. Student’s t test was performed to analyze the statistical significance between groups unless otherwise indicated, and p value < 0.05 was considered statistically significant.