The p52 isoform of SHC1 is a key driver of breast cancer initiation

All studies were performed on rats generated and maintained on the SS/JrHsdMcwi (SS) genetic background. Generation of the p66SHC-KO rats has been described previously [19]. Briefly, p52SHC knockout rats (SS-Shc1^em6Mcwi) were generated by pronuclear injection of a CRISPR plasmid expressing Cas9 and single-guide RNA targeting the sequence CACTCAGCTTGTTCATGTCCTGG (protospacer adjacent motif underlined) into one-cell SS (SS/JrHsdMcwi) rat embryos. Founder animals were genotyped by Cel-1 assay and confirmed by Sanger sequencing to identify a founder animal harboring a 6-bp deletion including the p52SHC initiation codon. All subsequent litters are genotyped using the following primers: p52ShcKO forward 5′-CCCTCCTCCAGGACAAGC-3′; wild-type forward 5′-TCCTCCAGGACATGAACAAGC-3′; common reverse 5′-TATGCACTCACCCGAACCAA-3′. p52SHC-KO rats are born in the expected genotypic ratio with normal litter size and able to breed as homozygous knockouts. All rats were fed ad libitum on a low salt (0.4% NaCl, Teklad7034) diet. Animal use and welfare procedures adhered to the NIH Guide for the Care and Use of Laboratory Animals following protocols reviewed and approved by the Medical College of Wisconsin Institutional Animal Care and Use Committee.

To induce mammary tumors, 45- to 55-day-old female rats (WT n = 14, p52SHC-KO n = 12, p66SHC-KO n = 13) were administered 65 mg/kg of body weight of 7,12-dimethylbenz(a)anthracene (DMBA) by oral gavage, as described [20]. Animals were then palpated weekly for tumor formation in the mammary fat pads. The latency of tumor development is defined as the time between DMBA administration and detection of the first tumor, whereas tumor multiplicity is the number of tumors per rat at week 15. At week 15 post-DMBA, rats were euthanized by CO₂ asphyxiation and number of tumors and total tumor burden per rat were recorded. Tumor burden is calculated as the sum total weight of all tumors from one animal.

Carmine alum staining of whole inguinal mammary glands was performed using standard protocols. Briefly, excised fourth and fifth mammary glands from the left and right side of adult approximately three-month-old rats were stretched and pressed onto a 2 × 3 in. glass slide and fixed 24 h in Carnoy’s fixative. Following rehydration to H₂O, slides were stained in Carmine Alum stain 24 h and then serially dehydrated in ethanol. Finally, slides were cleared in Histo-Clear™ clearing reagent up to 72 h until translucent and coverslipped using Histomount mounting media. Images were taken on a Nikon SMZ1500 microscope.

Quantitative real-time PCR was performed using Power SYBR™ Green PCR Master Mix (Invitrogen) on a TissueScan cancer survey cDNA panel (OriGene), covering 8 different cancers, with three normal tissue controls and 9 samples for each type of cancer included. Primers were validated for efficiency using cDNA from human mesangial cell cultures, known to express all three SHC protein isoforms. Additionally, dissociation curve analysis and gel electrophoresis of PCR products were used to validate specificity. Primer sequences were p52Shc/p46Shc forward 5′-GGCCCTGGACATGAACAA-3′ and reverse 5′-CCATGACTTTGTCGTTGGGA-3′; p66Shc forward 5′-CTGAAACTGTCTGGGTCTGAG-3′ and reverse 5′-TAGCCTGGTTGGACCTCT-3′; β-actin control primers were provided with the TissueScan kit and were used as a reference gene.

Total RNA was extracted by Trizol from whole tumors that were excised from WT (n = 5), p52SHC-KO (n = 5), and the p66SHC-KO (n = 5), followed by library preparation using Illumina’s TruSeq RNA library kit and sequencing on an Illumina HiSeq2500 (Illumina, Inc., San Diego, CA). Individual libraries were prepared for each sample, indexed for multiplexing, and then sequenced on an Illumina HiSeq2500. The Trim Galore program (v0.4.1) was used to trim bases with a Phred quality score < 20 [https://​www.​bioinformatics.​babraham.​ac.​uk/​projects/​trim_​galore/​] and only reads with a Phred quality score equal or higher than 20 were taken for analysis. The RSEM program function “rsem-prepare-reference” (v1.3.0) was used to extract the transcript sequences from the rat genome (build Rnor6.0) [26] and to generate Bowtie2 indices (Bowtie2 v2.2.8) [26], followed by read alignment using the “rsem-calculate-expression” function. Differential expression analysis was performed using the Bioconductor package DESeq2 version 1.12.4 to compute log2 fold changes and false discovery rate-adjusted P values [27]. Statistical significance was determined at a false discovery rate threshold of 0.05. Data were analyzed for molecular and functional pathway enrichment using Ingenuity Pathway Analysis (IPA; Qiagen, Redwood City, CA, USA).

Frozen OCT-embedded human breast tissue (tumor and matched normal) was obtained from the Medical College of Wisconsin Tissue Bank. For these specimens, normal tissue is defined as being distant from the tumor of the same patient that has no visual tumor present as determined by the pathologist at the time of procurement. Specimens were washed with PBS to remove OCT compound before protein extraction, performed using a Potter-Elvehjem homogenizer followed by sonication in RIPA buffer. Lysates were centrifuged to remove debris, and protein was measured using the BCA assay (Thermo Fisher) and frozen at − 80 °C before SDS-PAGE western blot analysis. SHC protein expression was analyzed using anti-SHC1 antibody (EMD Millipore 06-203) (dilution 1:2000) followed by IRDye800 secondary antibodies (Li-COR). Total protein was assessed by REVERT (Li-COR). Human INSTA-Blot™ Breast Tissue OncoPair (Novus Biologicals #NBP2-30128) ready-to-use PVDF membranes were used to assess total SHC proteins expressions in human breast cancer and normal adjacent tissue, 14 μg total protein per sample. Total protein was assessed by amido black stain provided by the manufacturer. Membrane was blotted with anti-SHC1 supplemented with anti-p66SHC antibody (Biosource 44829M) (dilution 1:1000) followed by goat anti-rabbit-HRP and goat anti-mouse-HRP (Biorad) (dilution 1:2000), according to manufacturer’s protocol. For rats, tumor and distant normal mammary samples from the same rat were also compared. Thirty micrograms of total protein was electrophoresed and transferred to Immobilon-FL membrane (Millipore). Membranes were blocked in Odyssey Blocking Buffer/TBST 1:1 and then blotted with anti-SHC1 antibody (EMD Millipore) (dilution 1:1000) and anti-β-actin (Sigma) (dilution 1:2000) overnight in 1:1 Odyssey Blocking Buffer/TBST. Membranes were rinsed in TBST and incubated with appropriate IRDye secondary antibodies (Li-COR) at dilution 1:15,000. Membranes were scanned on a LI-COR Odyssey scanner.

Staining for ER, PR, HER2, and SHC1 was performed using Dako EnVision FLEX mini Kit on a Dako Autostainer Omnis (Agilent, Santa Clara, CA). High-resolution digital images were captured at × 20 using a Pannoramic 250 Flash III slide scanner (3DHISTECH Ltd., Budapest, Hungary). ER, PR, HER2, and SHC1 were detected using EnVision FLEX Target Retrieval Solution Low pH (Citrate buffer, pH 6.1, Agilent) and incubation with anti-ER (1:200, ab3575, Abcam, Cambridge, MA), anti-PR (1: 2000, ab16661, Abcam), anti-HER2 (1:400, A0485, Agilent), and anti-SHC (1:50, 06-203, Sigma) for 30 min, followed by secondary antibody (Polymer) for 30 min. Counterstaining with hematoxylin, dehydration, clearing, and coverslipping was performed. The Dako HER2 antibody recognizes rat HER2 [28].

According to the Human Protein Atlas, in breast carcinomas, there is a difference in expected outcome between patients expressing high levels of SHC1 RNA and patients expressing low levels of SHC1 RNA based on best separation (P = 0.0412), but not median separation (P = 0.0928) [29]. Even though these data suggest the significance of high levels of SHC1 RNA for the worst expected outcome of breast cancer, they do not address the impact of individual protein isoforms in the progression of breast carcinomas. In mammals, SHC1 is expressed as two distinct transcripts based on the utilization of alternate promoters [9]. The shorter (3062 bp) transcript, termed here the p52SHC/p46SHC transcript, contains in-frame initiation codons for both p52SHC and p46SHC. The longer (3481 bp) transcript, termed the p66SHC transcript, contains initiation codons for all three protein isoforms, but is considered to serve predominately as a source of p66SHC. A broad human TissueScan cancer survey panel was screened for expression of both SHC1 transcripts. Each panel contained cDNA, normalized for β-actin expression, from 9 cancer samples from individual patients of increasing stage from 8 different cancer types and three normal tissues [30]. Expression of p52SHC/p46SHC transcript was upregulated more than 2-fold (*) compared to non-cancer tissue in 5 out of the 9 breast cancer samples, with one additional sample being greater than 1.5 times (#) higher (Fig. 1a) when compared to the median non-tumor samples. According to supplied information from OriGene, most of the tumors assayed are estrogen receptor (ER) and prolactin receptor (PR) positive, with mixed Her2 positive staining (Additional file 1: Figure S1). In contrast, no significant changes in p52SHC/p46SHC transcript were observed in colon, ovary, and thyroid cancers. Prostate, lung, kidney, and liver carcinomas exhibited variable expression in p52SHC/p46SHC expression, ranging between 1.5- and 2-fold increase over normal tissue (Fig. 1a). Breast carcinomas were unique in that most samples expressed the p52SHC/p46SHC transcript 2-fold higher than the normal breast tissues. Expression of the p66SHC transcript, the lesser-expressed Shc1 transcript, was below detectable levels in most normal and some breast tumor samples (data not shown). Additionally, p66SHC transcript levels were below detectable levels in many of the tumor samples across all cancer types (data not shown).

To assess differences in expression of all three SHC1 protein isoforms in human breast tumors, we have analyzed 9 independent breast cancer tumor samples (T) paired with matched normal protein samples (N) from tissues obtained from MCW Tissue Bank. It is of note, that in all (five) ER+/PR+/HER2− breast cancer samples, we detected the increased expression of p52SHC and p46SHC whereas the difference in p66SHC expression was not observed (Fig. 1b, c). The tumor/stroma composition (contribution of normal mammary epithelial) was quantified by a pathologist (Additional file 5: Table S1). Although triple negative tumors appeared to have increased SHC1 proteins levels, the differences did not reach significance compared to normal tissues when normalized to the total amount of protein (Fig. 1d, e). Additionally, commercial INSTA-Blot™ OncoPair membranes were blotted with total SHC1 antibody that recognizes all three isoforms, concurrently with an N-terminal p66Shc antibody. Comparison of seven independent tumor samples of increasing grade and stage [31] with matched normal protein samples from adjacent tissues indicated that p52SHC and p46SHC isoforms were upregulated in 6 out of 7 tumor samples compared to normal-paired samples (Fig. 1f). Likewise, 3 out of 7 tumor samples expressed increased levels of p66SHC protein compared with the normal-paired breast tissues. Similar to observations made with human breast tissue, the analysis of DMBA-induced rat tumors compared with normal rat mammary tissue revealed increased expression of all SHC1 isoforms (Fig. 1g). Immunohistochemical staining of SHC1 proteins showed markedly elevated expression of SHC1 proteins in breast cancer cells compared to adjacent normal breast epithelia or stromal cells, both in human (Fig. 2) and rat tumors (Additional file 2: Figure S2).

Individual contributions of p66SHC and p52SHC in the development and progression of breast cancer are unclear. Since both p66SHC and p52SHC protein levels were elevated in some human and in rat DMBA-induced tumors (Fig. 1d, e), we utilized rats in which p66SHC or p52SHC proteins were genetically removed to study the consequence of their loss on DMBA-induced mammary tumor formation. The p66SHC knockout (p66SHC-KO) was generated by targeting the unique sequence of p66Shc using Zinc Finger technologies and has been previously described [19]. For the first time in any species, a p52SHC knockout was generated using CRISPR/Cas-9 genomic editing [32]. A single founder was generated in which 6-base pairs, including the p52SHC ATG initiation codon, were deleted (termed p52SHC-KO, Fig. 3a). This deletion resulted in a 2-amino acid loss from p66SHC protein that was not predicted to alter p66SHC function. The deleted amino acids are not located in any known functional protein domains [12]. Furthermore, p52SHC-KO rats do not exhibit common phenotypes associated with p66SHC loss of function when placed on a high salt diet (data not shown), suggesting a normal p66SHC function in the p52SHC-KO rats. Additionally, the p46SHC sequence was preserved. Importantly, the mutation did not alter expression of either p66SHC or p46SHC protein levels, while p52SHC isoform lacks completely as detected by PCR genotyping (Fig. 3b) and western blot using a total SHC1 antibody (Fig. 3c, Additional file 3: Figure S3). Absence of p66SHC band by western blot analysis in p66SHC-KO, but not in WT or p52SHC-KO rats is also demonstrated (Additional file 3: Figure S3). Additionally, both p52SHC-KO and p66SHC-KO rats can be maintained at homozygous knockouts and give birth to average litter size and pup size (data not shown). There was no difference in the mammary gland structure between WT, p52SHC-KO, and p66SHC-KO rats (Fig. 3d). No masses, cysts or other lesions were identified across all three groups. Thus, changes in the development of mammary glands due to p52SHC or p66SHC ablation are likely not responsible for differences in DMBA-induced mammary tumor formation. Following administration of DMBA, loss of p52SHC resulted in an increase in tumor latency and a decrease in tumor incidence to 58.3% (7 of 12) by 15 weeks, compared to WT (92.9%, 13 of 14) and p66SHC-KO (100.0%, 13 of 13), (P = 0.038 and P = 0.003, respectively, Fig. 4a). Differences in tumor development were apparent as early as 10 weeks post-DMBA in a pair-wise comparison, 25.0% (3 of 12) p52SHC-KO rats had tumors compared to 78.6% (11 of 14) for WT and 76.9% (10 of 14) p66SHC-KO (P = 0.02 and P = 0.03 respectively). There was no difference in latency between WT and p66SHC-KO at any point during the 15-week time course (P = 0.4). During the 15 weeks that tumor progression was analyzed, rats were individually palpated for the presence of tumors. It may be possible that very small lesions were missed in the analysis. In addition to increased latency, p52SHC-KO rats also had lower tumor multiplicity, averaging 1.0 ± 0.3 tumors per rat compared to 2.75 ± 0.4 and 4.31 ± 0.3 tumors in WT and p66SHC-KO, respectively (P = 0.002 and P < 10⁻⁴, Fig. 4b). Furthermore, p52SHC-KO rats had a lower overall tumor burden per rat, 0.08 ± 0.04 g, compared to WT, 1.34 ± 0.5 g, and p66SHC-KO rats, 3.10 ± 0.8 g, respectively (P < 0.02 and P < 10⁻⁵, Fig. 4c). In contrast, p66SHC-KO rats trended towards increased multiplicity and tumor burden compared to WT DMBA-induced rats but was not significant (P = 0.09 and P = 0.08). We have carried out immunostaining for ER, PR, and HER2 of DMBA-induced mammary tumors formed in our genetically modified rats (Additional file 6: Table S2). All analyzed DMBA-induced tumors were ER and PR positive and HER2 negative.

To begin identifying gene networks that are mediated by p52SHC and p66SHC in breast cancer, RNA-seq was performed on mammary tumors derived from WT (n = 4), p52SHC-KO (n = 5), and p66SHC-KO (n = 5) rats. Compared with the WT tumors, 893 genes were differentially expressed (DE; FDR < 0.05) in the p52SHC-KO tumors, whereas only 18 DE genes were detected in the p66SHC-KO tumors (Fig. 5a, Additional file 7: Table S3). Although the WT and p66SHC-KO tumors were largely indistinguishable, gene ontology (GO) enrichment analysis using IPA revealed multiple enriched GOs in p52SHC-KO tumors compared with WT (Fig. 5b), including the ESR1 pathway (68 DE genes, Z-score = − 2.2, P = 1.46 × 10⁻³) and the RICTOR/mTORC2 pathway (46 DE genes, Z-score = − 6.8, P = 6.17 × 10⁻¹⁹). Collectively, these data provide the first in vivo link between p52SHC isoform and well-known drivers of breast cancer.