Background
Breast cancer is the second leading cause of cancer-related deaths in women worldwide [
1]. With an estimated 276,480 newly diagnosed cases in the USA alone, breast cancer is the most frequently diagnosed malignancy in women. Between 25 and 30% of these breast cancer patients develop metastases, at which point the disease becomes incurable, and more than 42,000 patients die each year. Clinically, breast cancer is categorized into three major subtypes: (1) hormone receptor-positive, which refers to the presence of the estrogen receptor alpha (ER, encoded by the gene
ESR1) and/or progesterone receptor (PR); (2) human epidermal growth factor receptor 2 (HER2)-positive, which refers to the overexpression of HER2 due to amplification of the
ERBB2 locus; and (3) triple negative breast cancer (TNBC), which lacks expression of the above-mentioned markers [
2]. The majority of breast cancers are hormone receptor-positive, with nearly 75% expressing ER [
1,
3]. While short-term survival rates of ER+ breast cancer patients are higher than in other subtypes, such as TNBC, more patients die from ER+ breast cancer than from any other subtype. Hence, there is an urgent need to discover and develop novel effective therapies for ER+ metastatic breast cancer (MBC).
Spreading of the tumor to distant organs, or metastasis, has fatal consequences and is the cause of most breast cancer deaths. While ER+ breast cancer can be treated with endocrine (hormone) therapy, metastatic tumors often develop endocrine resistance. The most commonly used endocrine therapies for ER+ MBC are selective estrogen receptor degraders (SERDs), selective estrogen receptor modulators, and aromatase inhibitors [
4], with aromatases being key enzymes for the conversion of androgens into estrogens. Once ER+ tumors have metastasized and become endocrine resistant, effective treatments are limited. One established mechanism of endocrine resistance that limits treatment options is the acquisition of mutations in
ESR1, which render the ER protein less dependent on estrogen for its function [
5,
6].
ESR1 mutations occur almost exclusively in metastatic disease that recurs after endocrine therapy, the most common
ESR1 mutations cause ER to retain an active conformation without estrogen binding [
5‐
7]. In addition, breast cancers with
ESR1 mutations can gain a basal-like gene expression profile and enhanced M2-like macrophage immune activation that is associated with poor prognosis [
8].
Reliable model systems that recapitulate patients’ disease are fundamental for drug development with the goal to achieve better translatability into the clinic. However, models of human ER+ MBC are limited. Human breast cancer cell lines including T47D [
9], MCF7 [
9], and MDA-MB-231 [
10] have been extensively studied, and have been engineered to have
ESR1 mutations. Others have generated breast cancer cell lines from patient-derived xenograft (PDX) models [
11]. In response to the need for models that better recapitulate patient tumors, represent the wide range of inter-patient heterogeneity, and are reflective of today’s treatment paradigms, however, several ER + PDX models have been developed [
12‐
15]. PDX models resemble the patient’s disease on a genomic, molecular, and cellular level [
12,
15‐
19], and are therefore valuable tools for research.
One caveat of PDX models is that they are grown in immunodeficient mouse strains, and therefore lack a functional immune system. It is well established that the tumor microenvironment is involved in various stages of solid tumor progression and metastasis [
20‐
23] and, importantly, can influence treatment outcome and drug resistance [
24‐
26]. In breast cancer, T cell infiltration is dynamic and heterogenous, and T cells are often excluded from tumors due to an immunosuppressive microenvironment, rendering immunotherapy largely ineffective [
27]. On the other hand, myeloid cells robustly infiltrate breast tumors and are important for pro-tumorigenic effects [
28]. Cells of the myeloid lineage, such as tumor-associated macrophages (TAMs), have been shown to play a fundamental role in disease progression. In breast cancer, the most dominant macrophage types are TAMs with a tumor-promoting “M2-like” phenotype [
29], and high TAM infiltration correlates with poor prognosis [
30,
31]. TAMs can negatively influence the efficacy of PARP inhibitors [
24] and, in ER+ human breast cancer cell lines, TAMs have been implicated in resistance to endocrine therapy [
32]. Taken together, these data underscore the importance of generating immune-humanized models for ER+ breast cancer to better define tumor-host interactions and improve patient outcome by uncovering strategies to make this disease more responsive to immunotherapy.
Immune-humanization of mice has been previously described [
33], and such mice have been utilized in xenograft studies using human breast cancer cell lines [
34‐
36] or TNBC PDXs [
37]. However, immune-humanized ER+ PDXs have not been yet been reported. Thus, we sought to generate an immune-humanized PDX model of ER+ disease for which treatments are urgently needed: endocrine-resistant MBC with a naturally occurring
ESR1 mutation. Importantly, we found that immune-humanization of this breast cancer subtype recapitulates the lymphocyte-sparse and myeloid-rich immune milieu of certain human ER+ breast tumors: over 15% of luminal B breast cancers have been reported to display a lymphocyte-depleted milieu [
38].
Methods
Experimental animals and health monitoring
All procedures were approved by the University of Utah Institutional Animal Care and Use Committee (IACUC). NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) or NSG-SGM3 mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl Tg(CMV-IL3,CSF2,KITLG)1Eav/MloySzJ) were purchased from Jackson Laboratories one week before the beginning of an experiment and received 4100 ppm Uniprim Diet (3980X, 2% maltodextrin, TD cat. #200733) from Envigo upon arrival. Two days after arrival and throughout the entire experiment, mice received acidified drinking water and were supplemented with gel diet 76A (Clear H2O, cat. # 72-07-5022). During the first 3 weeks of an experiment and in cases of weight loss, mice also received peanut butter as a supplement (Spread The Love, cat. #B06ZXZ3JPZ) and some mice received oral antibiotic treatment with 19 mg/kg/day baytril Baytril/Enrosite (enrofloxacin). Health was monitored closely and mice weights were measured daily during the first three weeks of experiments and at least 3 times per week until the experiment ended. All procedures were performed in a sterile hood.
Preparation of acidified and estrogen supplemented drinking water
To prepare acidified drinking water, the pH of sterile water was adjusted to 2.8–3.0, which was previously shown to prevent infections in experimental mice [
39]. For estrogen (E2) supplementation, estrogen was added to acidified drinking water as described before [
15] at a final concentration of 8 μg/ml. Estrogen water was prepared freshly on the day of usage and changed every 3–4 days. Estrogen plasma concentrations in NSG or NSG-SGM3 mice supplemented E2 supplemented acidified drinking water is shown in Additional file
1: Fig S1.
Busulfan treatment
Busulfan (Selleck Chemicals, cat. #S1692) was reconstituted fresh at 49 mg/ml in DMSO immediately before each usage. For complete dissolving, busulfan solution was heated to 55 °C, sonicated for 20 min and kept warm at 42 °C on a heat block. Prior to injection, busulfan was diluted in warm PBS to reduce DMSO injection to a maximum of 4% per injection. For myeloablation, mice were intraperitoneally injected with 20 mg/kg busulfan on 2 consecutive days.
Injection of human HSCs
CD34+ hematopoietic stem cells (HSCs) isolated from healthy donor bone marrow (BM) were purchased from AllCells. On the day of injection, cells were thawed in a 37 °C water bath and transferred into a 50 ml conical tube. 20 ml of warm culture media (DMEM/F12 supplemented with 10% fetal bovine serum, FBS) were slowly added dropwise into the cell suspension while gently shaking the tube. Cells were centrifuged (200 for 20 min at RT) and washed again in 20 ml of culture media. After centrifugation (200×g for 20 min at RT), cell pellet was resuspended in 0.5–1 ml of HBSS and cell concentration was determined using an automated cell counter (Countess II). Cells were diluted in HBSS accordingly and kept on ice until injection. 85,000 cells were intravenously (IV) injected into experimental mice in 100 ul HBSS using a 29 G × 12.7 mm (1/2″) needle (BD Biosciences, cat. # 324702).
Implantation of breast PDX tumor tissue into mice
HCI-013 PDX tumor fragments were implanted into the inguinal mammary fat pad (MFP) of experimental mice as previously described [
40]. Briefly, tumor fragments were thawed, washed three times in warm culture media (DMEM/F12 supplemented with 10% FBS) and kept on ice until implantation. Mice were anesthetized, the area of future incision was disinfected and PDX tumor fragments were implanted into the inguinal MFP. The fat pad posterior to the tumor implant including the area of the inguinal lymph node was cleared and the skin was closed with staples. Wound clips were removed 11–14 days post-surgery and tumors were measured twice a week using a digital caliper (9 mm, Braintree Scientific, Inc.). Tumor volume was calculated using the formula:
$${\text{Volume}}_{{{\text{Tumor}}}} = \left( {{\text{length}} \times {\text{width}}^{{2}} } \right)/{2}$$
Estrogen pellet implantation and monitoring in vivo estrogen levels
Beeswax pellets containing 0.2 mg or 0.4 mg of E2 were prepared as published previously [
15,
40]. To implant E2 pellets into mice, the mouse was anesthetized and prepared for surgery similar to the PDX tumor implantation protocol. After an incision was made into the skin, a dull surgical forcep was then used to introduce an E2 pellet subcutaneously, contralateral to the tumor implantation site between the skin and the peritoneal wall. The E2 pellet was inserted into the pocket and the wound was closed using wound staples. To monitor systemic estrogen levels, one blood drop was obtained from the submandibular vein and plasma was isolated by centrifugation (2000×
g, 15 min, 4 °C). Estrogen concentrations were determined in the plasma using the Mesoscale Discovery (MSD) Custom Steroid Human Panel assay according to manufacturer’s instructions. Estrogen plasma concentrations of NSG or NSG-SGM3 mice implanted with E2 pellets is shown in Additional file
1: Fig. S1.
Bilateral ovariectomy and generation of HCI-013EI
To generate HCI-013EI, 6–8 week old NSG mice were bilaterally ovariectomized (OVX) as previously described [
15]. Briefly, one hour before surgery mice were given buprenorphine (0.1–0.2 mg/kg). Then, the day of surgery mice were anesthetized, two dorsal incisions were made and both ovaries were removed. PDX tumor fragments were implanted and another dose of buprenorphine (0.1–0.2 mg/kg) was given 8 h after the first dose. Mice were treated with carprofen (5 mg/kg) once a day for the following three days post-surgery to minimize pain. After harvesting the tumor, tumor cells were cultured for two weeks in phenol red-free HBEC medium [
40] supplemented with charcoal-stripped FBS. Tumor cells were resuspended in 15 ul growth-factor reduced Matrigel per mouse, and re-implanted into the MFP in bilaterally-OVX NSG mice. No E2 supplementation was given.
Tissue harvesting and processing
All studies were performed according to IACUC-approved procedures with veterinary supervision. Tumors were harvested before exceeding the approved maximum size of 3 cm for studies requiring immune reconstitution, or earlier in cases when the mouse health would be compromised. On the day of necropsy, mice were euthanized and blood was immediately drawn by cardiac puncture with a syringe that was filled with 50 ul of acid-citrate-dextrose (ACD)(Sigma Aldrich cat. #C3821). Blood was transferred into EDTA vacuum tubes (VACUETTE K2-EDTA, Greiner cat. #454052), which were also filled with 50 ul ACD, inverted twice and kept on ice. To isolate white blood cells, the blood was transferred into 15 ml conical tubes and red blood cells (RBCs) were lysed by adding 10 ml of 1 × RBC lysis buffer (Biolegend, cat. #420301) and incubating for 10 min at RT. Cells were spun down and RBC lysis was repeated if necessary. White blood cells were immediately stained for flow cytometry.
Tumors and MFPs were resected. Tumors were weighed and both tumors and MFPs were fixed in 10% NBF (neutral buffer formalin) for 24 h at 4 °C to generate formalin-fixed paraffin-embedded (FFPE) specimen.
Spleens were harvested and 1/5
th of the organ was fixed in 10% NBF 24 h at 4 °C to generate FFPE specimen. The remaining spleen was stored in DMEM/F12 media supplemented with 10% FBS and kept on ice. To generate a single cell suspension, spleens were placed onto a 70 µm cell strainer (Fisher Scientific, cat. #22363548) and mashed through it using the plunger of a 5 ml syringe (BD, cat. #309630). After rinsing the plunger and strainer with additional media, splenocytes were centrifuged (300×
g, 5 min, 4 °C) and RBC lysis was performed on the cell pellet at least twice as described above. Splenocytes were filtered again through a 70 µm cell strainer and directly stained for flow cytometry. For BM isolations, femur and tibia were removed and carefully cleaned from soft tissue using a scalpel. Femurs and tibiae were separated and BM was isolated by either flushing or centrifugation. To flush out the BM, both ends of the femurs and tibiae were cut off, and a 1 ml syringe (BD, cat. #309659) with a 26G needle (BD, 26G × 3/8, 0.45 mm × 10 mm, cat. #305110) was used to scour the BM. To centrifugate BMs, the knees were removed from the leg bones by cutting right above and below the knee, as previously published [
41]. Both tibia and femur were then placed into a 0.5 ml microtube with the bottom part cut off. The microtube containing the bones was then placed into a 1.5 ml microtube and centrifuged (≥ 10,000×
g, 15 s, RT). Isolated BM was resuspended and carefully broken into single cells using a pipette. Then the cell suspension was filtered through a 70 µm cell strainer, RBC lysis was performed, and cells were stained for flow cytometry.
All FFPE tissues were fixed in 10% NBF for 24 h at 4 °C, washed three times in PBS for 5 min and placed in 70% ethanol at 4 °C until tissue processing and paraffin embedding were performed. For all studies involving PDX tumor bearing animals, mice were checked for metastases in other organs, such as lungs and liver. If present, axillary lymph nodes were harvested and fixed as FFPE specimen as described above.
Flow cytometry
All flow cytometry analysis was performed using freshly isolated tissues and cells. Fc receptor blocking solution was prepared freshly by mixing a ratio of 10 μl/107 cells of mouse Fc receptor blocking (Miltenyi Biotec, cat. #130-092-575) with 15 μl/107 cells of human Fc receptor blocking (Miltenyi Biotec, cat. #130–059-901), then diluted in the same volume of FACS buffer (2% FBS in PBS). Cells were resuspended in 50 μl of Fc receptor blocking solution, transferred into a 96-well round-bottom plate (Greiner bio-one, cat. #650101) and incubated for 20 min on ice. The antibody cocktail was prepared by mixing the following antibodies: 0.5 μl/test BUV395 rat anti-mouse CD45 Clone 30-F11 (RUO) (BD Bioscience, cat. #564279), 5 μl/test APC mouse anti-human CD45 Clone HI30 (eBioscience, cat. #17-0459-42), 5 μl/test BV510 mouse anti-human CD11b/MAC-1 Clone ICRF44 (BD Bioscience, cat. #563088), 2.5 μl/test PE mouse anti-human CD56 (NCAM) Clone 5.1H11 (Biolegend, cat. #362508), 5 μl/test PerCP/Cy5.5 mouse anti-human CD3 Clone UCHT1 (Biolegend, cat. #300430), 5 μl/test FITC mouse anti-human CD19 Clone HIB19 (Biolegend, cat. #302206). The antibody mix of 23 μl/test was diluted with 27 μl/test FACS buffer. 50 μl of diluted antibody cocktail was added directly to each well and incubated for 30 min in the dark on ice. Unstained control wells received 50 μl of FACS buffer. After incubation, 100 μl FACS buffer was added to each well and the plate was centrifugated (300×g, 5 min, 4 °C). Cell pellets were washed once with 200 μl FACS buffer and once with 200 μl PBS. After the last centrifugation (300×g, 5 min, 4 °C), 100 μl of a 1:2000 dilution of Fixable Viability Dye eFluor® 780 (eBioscience, cat # 65-0865-18) in PBS was added to each well and incubated for 20 min in the dark. Unstained control wells received 100 μl of PBS. After incubation, 100 μl FACS buffer was added to each well and the plate was centrifuged (300×g, 5 min, 4 °C). Cell pellets were resuspended in 200 μl FACS buffer, transferred into a FACS tube (Olympus Plastics, cat. # 14-360) and stored at 4 °C with an aluminum foil cover until data acquisition.
Flow cytometry was performed immediately on freshly stained unfixed samples, and 0.5–1 million events/sample were acquired. Compensation was performed for each antibody individually using a 1:3 dilution of UltraComp eBeads™ Compensation Beads (Thermo Fisher, cat. # 01-2222-42) in PBS. 1 drop of the bead dilution was mixed with 1 μl of antibody and incubated for 20 min on ice in the dark. After centrifugation (600×
g, 5 min, 4 °C), supernatant was discarded, beads were resuspended in 200 μl PBS and 5000 events were acquired. FlowJo version 10.5.3 was used for data analysis. All flow cytometry experiments included controls (FMOs,fluorescent minus one), unstained cell, and human peripheral blood mononuclear cells (PBMCs) mixed with NSG organs as positive controls for all human antibodies. Cells killed by repeated freeze/thaw cycles were used as a control for the viability staining. All gates were set based on FMO controls and the gating strategy which was applied to all samples is displayed in Additional file
1: Fig. S2.
PDX tumor histology, IHC and IF stains
FFPE specimens were sectioned by the Biorepository and Molecular Pathology Shared Resource at Huntsman Cancer Institute. To obtain serial sections of tumors, 5 μm thick sections were collected, and hematoxylin–eosin (HE) staining was performed on sections every 35 μm to confirm tumor content and assess morphology.
All antibodies and immunohistochemistry (IHC)/immunofluorescence (IF) staining conditions used in this study are listed in Additional file
1: Fig. S3. Briefly, sections were incubated at 60 °C for 60 min, deparaffinized and rehydrated following standard procedures. Antigen retrieval was performed for 20 min in boiling buffer in the microwave, with the following exceptions. For PHH3 and CC3, the water bath was used at 60 °C ON, and for CAM5.2 ice-cold trypsin enzymatic antigen retrieval (abcam, cat. #ab970) was applied for 10 min at RT. For IHC (except for PHH3 and CC3), tissue slides were subsequently incubated in 3% H
20
2 methanol buffer for 10 min at RT to block endogenous peroxidase activity. Tissue sections were then incubated in PBS blocking buffer containing 5% bovine serum albumin (Sigma-Aldrich, cat. #A7906), 10% normal goat serum (Jackson ImmunoResearch, cat. #005-000-121), 10% normal human serum (Jackson ImmunoResearch, cat. #009-000-121) and FcR blocking reagents for mouse (1:100, Miltenyi Biotech, cat. #130-092-575) and human (1:100, Miltenyi Biotech, cat. #130-059-901) for 1 h at RT. In addition, if mouse primary antibodies were used, the M.O.M Immunodetection kit (Vector Labs, cat. #BMK-2202) was added to the buffer. For double IHC and IF, the Avidin/Biotin (Vector Labs, cat. #SP-2001) and Streptavidin/Biotin (Vector Labs, cat. #SP-2002) blocking kits were used, respectively. For single IHC, staining was visualized with the 3,3-diaminobenzidine—Peroxidase substrate (Vector Labs, cat. #SK-4100), using Hematoxylin as a counterstain (Sigma-Aldrich, cat. #MSH32). For dual IHC, VECTASTAIN ABC kit—Peroxidase HRP (Vector Labs, cat. #PK-4000) and Vector AEC—Peroxidase substrate (Vector Labs, cat. #SK-42000) were used for CAM5.2 visualization, whereas VECTASTAIN ABC-AP kit—Alkaline Phosphatase (Vector Labs, cat. #AK-5000) and Vector Blue—Alkaline Phosphatase Substrate (Vector Labs, cat. #SK-5300) were used for PHH3 and CC3 visualization. For double IF, Fluorescent Streptavidin kit (Vector Labs, cat. #SA-1200) was used to amplify hCD45 signal. Single and double IF sections were counter-stained with the nuclear stain 4′, 6-diamidino-2-phenylindole dihydro-chloride (DAPI) for 20 min at RT. IHC stains were mounted with the Cytoseal Mounting Medium, Richard-Allan Scientific, Iow (60 s) (VWR, cat. #48212-154) and the VectaMount™ AQ Mounting Medium (Vector Labs, cat. #H-5501), respectively. IF stains were mounted using the VECTASHIELD Vibrance Antifade Mounting Medium (Vector Labs, cat. #H-1700). For each staining, secondary antibody controls were performed on serial sections.
Slide digitalization and quantification
IF and IHC tissue slides were digitalized using a
Pannoramic MIDI II (3DHISTECH Ltd., Budapest, Hungary) and an Axio Scan.Z1 (ZEISS, Jena, Germany) automatic slide scanners, respectively. Automated staining quantification was done using QuPath software v0.2.3 [
42]. Briefly, after uploading the images and selecting for the appropriate image type (i.e. Brightfield: H-DAB for IHC or Fluorescence for IF) (Additional file
1: Fig. S4, step 1), the brightness/contrast tool was used on the entire image to improve the visibility of the stains. Subsequently, a small annotation was drawn (Additional file
1: Fig. S4, step 2) to train the software for positive cell detection by editing the appropriate parameters based on the cellular location and the intensity of each stain (Additional file
1: Fig. S4, step 3). Following the creation of a region of interest (ROI) encompassing the tumor and excluding the surrounding MFP (Additional file
1: Fig. S4, step 4), the improved positive cell detection analysis was finally applied to the ROI. Extra-tumoral tissue was excluded from the analysis in order to capture intratumoral immune cells only. Snapshots and images were taken with the CaseViewer software (2.4.0.119028). For each stain, no visible signal was detected in the secondary antibody control slides.
Droplet digital PCR to identify ESR1 mutations
Droplet digital PCR (ddPCR) was performed as previously published [
15,
43] to identify potential
ESR1 hotspot mutations in HCI-013 and HCI-013EI tumors. Specifically, RLT buffer (Qiagen, cat. # 1053393) containing beta-mercaptoethanol was added to tumor tissues which were then homogenized on ice. Immediately after, the QiagenAll Prep kit (Qiagen, cat. # 80204) was used to isolate genomic DNA (gDNA) and total RNA following manufacturer's instructions. After cDNA synthesis using PrimeScript RT Reagent Kit (Takara, RR037), 50 ng gDNA or cDNA was used as templates and primer/probe sets for four specific
ESR1 mutations (Y537S/Y537C/Y537N/D538G) was added [
43]. Droplets were generated using the Bio-Rad QX200 AutoDG Droplet Digital PCR system, followed by an amplification step of the
ESR1 ligand binding domain fragment using a thermo cycler. Wildtype (WT) and mutant probes were detected with a Bio-Rad QX200 droplet reader and the allele frequency of mutant
ESR1 was calculated using the QuantaSoft software version 1.7. To ensure calls were made correctly, droplets containing a positive control (gDNA from genome-edited Y537S
ESR1 mutant MCF7 cell line) or negative control (gDNA from
ESR1 WT MCF7 cell line), and background control (ddH
2O) were included.
AAVs for delivery of human cytokines
Adeno-associated virus 9 (AAV9) vectors encoding human interleukin-7 (IL7), IL15, thrombopoietin, or empty vector controls were purchased from Vector Biosystems Inc, diluted in PBS and IV injected into mice at the following concentrations: 1.0 × 1011 gc/ml AAV9-CAG-h-IL7-WPRE; 2.0 × 1011 gc/ml AAV9-CAG-h-IL15-WPRE; 1.0 × 1010 gc/ml AAV9-CAG-h-THPO-WPRE.
Discussion
We report for the first time the development of an immune-humanized PDX model of endocrine resistant ER+ MBC that harbors a naturally occurring
ESR1 Y537S mutation. Y537S has been shown to be the second most frequently occurring
ESR1 mutation in endocrine resistant breast tumors in patients and mutated
ESR1 leads to aberrantly increased ER transcriptional activity [
5‐
7]. To date, there are no treatments available that specifically target mutant ER
. This is of high relevance since patients whose tumors harbour
ESR1 mutations have poor prognosis compared to patients whose tumors are WT
ESR1 [
8]. Efforts have been made to generate valuable PDX models for ER
+ MBC, but the need to use immunodeficient hosts to grow PDXs makes it impossible to study the known role of the immune system in tumor progression and treatment response. Hence, we set out to use a novel approach for human immune reconstitution in our model. One of the major differences between our humanization approach and other humanized models is that our tumors are inoculated into mice 24 h after HSC injection. This enables human immune cells to be exposed to tumor antigens while they are developing in the mouse. Immune-humanization had no negative effect on tumor growth in our study (Fig.
6d), and we did not find a correlation between the degree of immune reconstitution and tumor growth rates, which are naturally heterogeneous in PDX models. We suggest that this strategy might reduce the risk of allogenic T cell responses, while also avoiding GvHD that could occur if patient-matched mature T cells are used for immune reconstitution.
Characterization of the tumor microenvironment of our immune-humanized model showed a predominant myeloid cell infiltration into tumors. CD11b+ myeloid cells are one of the most abundant immune cell component of the human breast tumor microenvironment, where they have been reported to interact with other factors and various tumor cells to foster cancer progression and metastasis [
61,
62]. Specifically, TAMs, myeloid-derived suppressor cells and dendritic cells have been shown to play a crucial role in tumor immune evasion by secreting IL-10, IL-12 and arginase-1, leading to suppression of CD8+ T-cell activity and reduced activity of effector T cells [
63,
64]. Interestingly, the receptor tyrosine kinase RON, which is expressed on resident macrophages, has been recently shown to promote endocrine therapy resistance in
ESR1 mutated breast cancers [
65]. Thus, since most of the tumor-infiltrating cells are of myeloid origin, our immune-humanized model would be an interesting system to study this mechanism further.
While we do detect human T cells (hCD3+) in peripheral blood and spleens of immune-humanized HCI-013EI PDX mice, T cells were not observed in the tumor microenvironment. Based on data from immune-humanized mice employing the use of HSCs, myeloid cells were shown to develop 8–9 weeks post-humanization, B cells develop after 10–11 weeks, while T cells develop after 1112 weeks, depending on the model used [
66,
67]. Accordingly, in experiments that lasted 14 or 18.5 weeks post-humanization, we detected more T cells in blood, BM and spleen than we did at 12 weeks post-humanization. Hence, one explanation for our observation could be that at the 12-week timepoint, when we needed to end the experiment with the HCI-013EI model due to tumor burden, T cells had developed but had not yet started to infiltrate the tumors. Alternatively, since we used NSG-SGM3 mice, which enable optimal myeloid immune reconstitution in mice [
44‐
46], it is possible that the myeloid cells restrict T cell recruitment into tumors, similar to what is seen in human breast cancers [
28]. Yet another possibility for lack of intratumoral T cells in our model is that HCI-013EI might represent a “cold” tumor. Cold tumors (i.e. noninflamed) are described as solid cancers that have low immune cell infiltration, including low or absent numbers of CD3+ and CD8+ T cells [
68‐
70]. Many breast tumors are considered to be immune “cold” [
38], and this is one of the reasons why immune checkpoint blockade is not very effective against this disease [
71]. In the case of HCI-013, it is impossible to know whether the original patient tumor was of the immune “cold” nature because no primary tumor material is available; the model was generated from a metastatic pleural effusion fluid sample.
Multiple efforts have been made to find ways to manipulate and improve the infiltration of T cells into the tumor. In one of these recent studies, vitamin D was shown to increase intratumoral T cells in a breast cancer model [
72]. In another study in melanoma patients, injection of oncolytic viruses led to an increase in CD8+ T cell tumor infiltration, thereby promoting the efficacy of anti-PD1 immunotherapy [
73,
74]. One possible application for the model we describe here would be to test novel strategies to increase T cell infiltration into the tumors as a way to increase susceptibility to immunotherapy for breast cancer.
The variable effects of immune-humanization in experimental mice that can be attributed to different HSC donors has been discussed at length, and various studies reported that the extent of humanization, as well as the development of individual immune subtypes, can vary from donor to donor [
75,
76]. Using different HSC donors, we found similar trends in the timing of human immune cell development, with myeloid cells being the predominant immune cells detected at earlier timepoints post-humanization, while T cells developing at later stages. We also found that E2-induced anemia of immune-humanized mice was consistent across multiple HSC donors. Indeed, the veterinary literature contains several reports of estrogen-mediated BM depression followed by severe anemia in other species including dogs and ferrets [
77,
78]. This phenomenon underscores a barrier to the development of immune-humanized models of E2-dependent cancers.
Lastly, the time required for the differentiation of various immune subsets differs and also depends on cytokine signaling. NSG-SGM3 mice are engineered to provide human cytokines needed for myeloid development, but the complex milieu of cytokines and other factors necessary for the development of a complete immune system is still lacking. As proof-of-concept, we noted that we could significantly alter immune cell differentiation by delivering additional exogenous cytokines. Using AAVs as vectors to deliver human TPO, IL-7 and IL-15 to NSG-SGM3 mice in our studies, we found that we could strongly skew immune development towards NK cells at the expense of other cell types (Additional file
1: Fig S6a–f). The addition of cytokines or AAVs did not affect tumor size in two different experiments, and we did not find an effect of matching killer immunoglobulin-like receptors (KIR) between HSC donors and the tumor (Additional file
1: Fig. S6g). This study emphasizes the importance of how a balanced supply of human immune cytokines may be tailored to influence development of immune cell types of interest. Future work will be required to fine-tune physiologically-relevant levels of various cytokines and recapitulate the different immune subtypes that dominate in breast cancer [
38].
Acknowledgements
We would like to thank the University of Utah Office of Comparative Medicine for animal care. We also want to thank Andrew J. Butterfield, Shanna Kuhn, other members of the Welm Lab, and the Huntsman Cancer Institute Preclinical Research Resource for assistance in harvesting and processing organs, and flow cytometry. This work was supported by the DOD Breast Cancer Research Program Era of Hope Expansion award W81XWH-12-1-0077 (to ALW), NCI R01CA221303 (SO), as well as pilot funds from the Utah Center for Clinical and Translational Science, funded by the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number UL1TR002538. SERD experiments were sponsored by Zentalis Pharmaceutical, Inc. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The University of Utah Flow Cytometry Facility is supported in part by 5P30CA042014-24 from the National Cancer Institute.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.