Tissue procurement and IRB approval including patient consent were obtained from the Cooperative Human Tissue Network.

Skin and adipose tissue was obtained from patients undergoing cosmetic surgery. Adipose tissue was incubated with Hanks' balanced salt solution (HBSS), 1% fetal bovine serum, 0.04% sodium bicarbonate, 1% HEPES, 0.5% amphotericin B, 1% streptomycin/penicillin and 0.1% collagenase type 1A. Cells were placed into a 37°C orbital shaker for 1 h with constant agitation at 4 × g. The cell slurry was centrifuged at 360 × g for 15 min and adipocytes washed, suspended in medium (DMEM with 4.5 g/L glucose, 10% fetal bovine serum, 1% streptomycin/penicillin, 1% L-glutamine and 1% HEPES), and cultured for up to 7 days or until the stem cells were confluent before testing of MC-differentiating media below.

Different cell culture media were tested for their ability to induce MC differentiation of the adipose cells using X-VIVO 15 or AIM-V (Lonza, Switzerland), plus 80 ng/ml SCF (Stemcell Technologies, Vancouver, BC) with or without non-specific (NS) psIgE (human myeloma IgE; a gift from Dr. Andrew Saxon, UCLA; 0.1 μg/ml). Conditioned MC media was produced using media used to culture primary human skin MC as described (12, 25). Briefly, skin MC cultures (>5 weeks) containing 80 ng/ml SCF in X-VIVO 15 were pelleted by centrifugation, supernatants removed, filtered through a 22 μm filter (Sigma-Aldrich, St. Louis, MO) to remove any cells, and added directly to the adipose stem cells (~15 ml per 75/mm² flask). Approximately every seven to 10 days, viability was assessed by trypan blue exclusion and half of the media was collected and replaced with fresh media. Initial monitoring of MC differentiation was determined using toluidine blue staining of cytospins followed by further characterization as described below.

Flow cytometry was performed using a FACS Arial III (Becton Dickenson, Franklin Lakes, NJ). Briefly, mouse anti-human Abs to FcεRI, c-kit, FcγRI, FcγRII, FcγRIII (Santa Cruz, Dallas, TX), or mouse IgG isotype control MOPC (Sigma-Aldrich) were added for at least 1 h on ice, washed, and F(ab′)₂-FITC-goat anti-mouse Abs (BD Biosciences, San Jose, CA) added for detection (26). All experiments were performed at least three times.

Immunochemistry was performed with mouse anti-human Abs to tryptase and chymase or NS mouse IgG isotype (negative) control as described (27, 28) but using Cy3-conjugated anti-mouse secondary Abs. For detection of ADMC-induced apoptosis of human breast cancer SK-BR-3 cells (ATCC, Manassas, VA), cell cytospins were incubated with 1 μg/ml Alexa Fluor^TM 488 dye (ThermoFisher Scientific, Walnut, CA) labeled mouse anti-human tryptase (1 μg/ml; for ADMC detection; green) along with Alexa Fluor^TM 647 labeled mouse anti-human Ab to the active form of caspase 3 (1 μg/ml; for SK-BR-3 detection; red) or Alexa Fluor^TM 647 labeled isotype control for the caspase 3 Ab. To quantify the percentage of caspase 3 positive cells observed on the cytospins a total of 200 cells were counted on each slide and the number of SK-BR-3 cells positive for caspase activation was compared to the number of those not stained for caspase 3 that were not MC.

RNA was extracted from ADMC using the Qiagen RNeasy Plus Mini kit (Qiagen, Germany). Reverse Transcriptase PCR (RT-PCR) was performed using the Qiagen OneStep RT-PCR kit using primers previously described to amplify short fragments of the β-actin, tryptase, chymase, c-KIT, and FcεR1α RNA (29). Cycling conditions were: 50°C for 30 min, 95°C for 15 min, followed by 35 cycles of 94°C for 45 s, 53–63°C for 45 s (according to primer T_m), 72°C for 1 min and a final 10 min extension at 72°C.

The fully human anti-human HER2/neu IgE/kappa containing the variable regions of the human scFv C6MH3-B1 has been previously described (30). In addition, we also developed an anti-human HER2/neu IgE/kappa containing the variable regions of the humanized Ab trastuzumab (Herceptin®) by subcloning the variable regions of trastuzumab previously used in Ab-cytokine fusion proteins (31, 32) into the human epsilon/kappa expression vectors use to the develop the C6MH3-B1 IgE. The trastuzumab IgE and C6MH3-B1 IgE bind different epitopes of human HER2/neu. They are expressed in murine myeloma cells and the transfectomas grown in roller bottles for Ab production as described (30). The IgE Abs are purified from cell culture supernatants on an immunoaffinity column prepared with omalizumab (Xolair®) (Genentech, Inc. San Fransisco, CA, USA) (30). The extracellular domain of HER2/neu (ECD^HER2) was produced as described previously (31). All proteins were quantified using the BCA Protein Assay (ThermoFisher Scientific).

To determine ADMC functional responses mediated through FcεRI, ADMC were incubated with 1 μg/ml of anti-FcεRI Abs or with 1 μg/ml anti-NP IgE for 1 h followed by NP-BSA. To determine ADMC functional responses mediated by non-IgE pathways, ADMC were incubated with 40 μg/ml Poly-L-Lysine (Sigma-Aldrich) or 10 μM A23187 (Sigma-Aldrich). Post-incubation, activation was performed for 30 min (to measure degranulation) or overnight (for cytokine analysis) and β-hexosaminidase release and TNF-α and GM-CSF production were measured as described (33–35). All experiments were performed in duplicate from four separate donors and significant differences (p < 0.05) determined using the Student t-test.

To assess the ability of anti-HER2/neu IgE sensitized ADMC to bind to HER2/neu expressing SK-BR-3 breast cancer cells, confocal imaging was used on differentially labeled, live cells. The ADMC (1.5 × 10⁵) were sensitized with 1 μg/ml of anti-HER2/neu IgE Abs or NS psIgE followed by the addition of MitoTracker™ Green (500 nM; ThermoFisher Scientific). The ADMC were washed once in warm X-VIVO 15 and added to the adherent, human HER2/neu-positive SK-BR-3 cells that were pre-stained with MitoTracker™ Red (500 nM; ThermoFisher Scientific) in a live cell incubator affixed to a confocal microscope and images acquired over 6 h.

ADMC were sensitized with or without 1 μg/ml of anti-HER2/neu IgE or NS psIgE as above and added to human breast cancer cells expressing high levels of HER2/neu SK-BR-3 or BT-474 (a gift from Dr. Hui-Wen Lo, Wake Forest University) cells for 1 h in 24 well plates. The ratio of MC to breast cancer cells varied from 1:10 to 10:1 ADMC to breast cancer cells and mediators assessed in the supernatants. In some experiments anti-HER2/neu IgE sensitized ADMC challenged with ECD^HER2 or heat-inactivated serum from patients with HER/neu positive breast cancer (Cureline, Brisbane, CA; Table 1).

Three different methods were used to assess the ability of anti-HER2/neu IgE sensitized ADMC to induce cell death of HER2/neu expressing breast cancer cells. First, ADMC (1.5 × 10⁵) were sensitized with 1 μg/ml of anti-HER2/neu IgE or psIgE for 2 h. Breast cancer cells (5 × 10⁴) on coverslips were labeled with 2 μM MitoTracker™ Green (ThermoFisher Scientific) for 1 h. The washed ADMC were labeled with CellTracker^TM Deep Red (which stains the cells reddish/purple under confocal; 2 μM) for 1 h, washed, and added to SK-BR-3 in medium containing 25 μg/ml of propidium iodide (PI; which stains the cells red) used to detect dead cells (36) and PI intensity measured over time. Second, SK-BR-3 were plated and incubated with Cellevent^TM Caspase 3/7 Green (to detect activated caspase-3/7 in apoptotic cells; ThermoFisher Scientific) for 1 h according to the manufacturers protocol. ADMC, treated with MitoTracker^TM Red (1 μg/ml), were added to the washed SK-BR-3 cells and incubated for up to 4 days. Third, cytospins of cells from separate experiments were made and used for immunofluorescence detection of apoptosis. Briefly, cytospins were fixed in methanol and incubated with Alexa Fluor^TM 488 dye (ThermoFisher Scientific) labeled mouse anti-human tryptase (1 μg/ml; for ADMC detection; green, for co-cultures) along with Alexa Fluor^TM 647 labeled mouse anti-human active caspase 3 (BD Biosciences, 1 μg/ml; for SK-BR-3 detection; red) Alexa Fluor^TM 647 labeled isotype Abs were used as a control for the caspase 3 Ab.

In separate experiments, cell free supernatants from optimally activated ADMC (1.3 × 10⁶) by an anti-FcεRI Ab (1 μg/ml for 24 h; 60–70% release) were directly added to MitoTracker™ Green-labeled SK-BR-3 cells (5 × 10⁴). In some experiments, an anti-human TNF-α Ab (5 μg/ml) was added to the supernatants to block TNF-α activity. Cell death was monitored over time through the quantification caspase 3/7-positive cells (>200) counted at the end of each experiment to obtain percentages. All confocal/live cell experiments were performed on three separate ADMC donors and significance (p < 0.05) tested using the Student's t-test.

Several culture conditions were tested for their ability to induce MC differentiation (35, 37). The conditioned media from skin-derived human MC cultures was found to be optimal for ADMC differentiation (Table 2). In conditioned medium, ADMC were observed to emerge from large clumps of cells or tissue as shown in Figure 1A. After 3–4 weeks of culture, mature MC (>90% viable) were observed as demonstrated by the classical spherical, highly granulated morphology (Figure 1B) characteristic of skin-derived MC (Figure 1C). In addition, the ADMC were positive for messenger RNA to the two major MC proteases, tryptase, and chymase (Figure 1D). Furthermore, the ADMC expressed surface markers for tissue MC including FcεRI and the receptor for SCF, c-kit (Figure 1E). As previously reported with skin MC (26), ADMC express FcγRII and not FcγRI or FcγRIII (Figure 1F). As seen in Figure 1G both tryptase and chymase protein was detected using immunohistochemistry. Thus, adipose tissue has MC progenitors that can be differentiated into MC that are phenotypically similar to human connective tissue (MC_TC) (38) based on these characteristics. Representative numbers of ADMC obtained from surgical specimens are shown in Table 3.

The functional response of ADMC was compared to skin-derived MC. As seen in Figure 2, ADMC degranulated (Figure 2A) and produced cytokines (Figure 2B) in response to FcεRI engagement. Cytokine release by ADMC and skin MC was similar in response to FcεRI-dependent stimuli averaging 2,850 and 2,600 pg/ml of GM-CSF in skin MC and ADMC, respectively. A similar degranulatory response with ADMC was observed using non-FcεRI-dependent stimuli Poly-L-Lysine and A23187 (Supplemental Figure 1). Taken together, the ADMC are functionally similar to skin-derived MC in response to FcεRI-dependent and FcεRI-independent stimuli.

The ability of ADMC sensitized with the anti-HER2/neu IgE to bind HER2/neu–positive SK-BR-3 breast cancer cells was investigated. As seen in Figure 3A, the ADMC sensitized with the anti-HER2/neu IgE (green) bound to HER2/neu-positive SK-BR-3 breast cancer cells (red) as demonstrated in the time lapse pictures and video (Supplemental Video 1). However, ADMC sensitized with a NS IgE did not target or bind to the SK-BR-3 cells (Figure 3B). These results demonstrate that the anti-HER2/neu-IgE mediates the interaction between ADMC and HER2/neu-positive breast cancer cells.

ADMC must release their mediators upon FcεRI challenge at the site of the tumor to be effective anti-tumor agents. Thus, the ability of ADMC sensitized with the anti-HER2/neu IgE to degranulate in the presence of breast cancer cells was investigated. ADMC were sensitized with one of two anti-HER2/neu IgE Abs recognizing different epitopes (trastuzumab IgE or C6MH3-B1 IgE). Varying ADMC cell numbers were incubated with SK-BR-3 breast cancer cells and mediator release assessed in the medium. As seen in Figure 4, ADMC sensitized with anti-HER2/neu IgE induced significant (p < 0.05) mediator release through FcεRI when co-incubated with the HER2/neu-positive SK-BR-3 breast cancer cells. The ADMC degranulated to release pre-stored mediators (Figure 4A), as well as newly formed mediators TNF-α and GM-CSF (Figure 4B). Another HER2/neu-positive breast cancer cell line, BT-474, also induced degranulation and cytokine production optimally at a ratio of 1:2 (degranulation) and 1:0.5 (cytokine release) ADMC:BT-474 (Supplemental Figure 2).

The above results suggest the possibility of using ADMC armed with IgE Abs can trigger degranulation in the presence of HER2/neu expressing cancer cells and thus, the potential of using this strategy for cancer therapy via the release of MC mediators. However, a potential concern of the systemic administration of ADMC sensitized with an anti-HER2/neu IgE is the possible induction of a systemic anaphylactic reaction as patients with HER2/neu breast cancer can have elevated levels of circulating ECD^HER2 in the blood (39, 40). The IgE Abs are not expected to induce FcεRI cross-linking when complexed with soluble antigen (ECD^HER2), given the mono-epitopic nature of this interaction and the fact that ECD^HER2 does not form homodimers in solution (41, 42). To address this concern, the ability of ECD^HER2 to induce FcεRI-mediator release was examined. As described previously (30), ECD^HER2 in the presence of the anti-HER2/neu IgE Abs did not induce degranulation, while anti-FcεRI Abs induced release (Figure 4C). Furthermore, serum from two separate HER2/neu positive breast cancer patients did not induce ADMC degranulation (Figure 4D). These results suggest that the anti-HER2/neu IgE-sensitized ADMC will not induce a systemic anaphylactic response in vivo and will only release mediators upon encountering HER2/neu on breast cancer cells.

The ability of ADMC to induce breast cancer cell death was investigated. ADMC sensitized with the anti-HER2/neu IgE were added to SK-BR-3 cells in medium containing PI to discriminate dead cells from live cells (36). As seen in Figure 5A, binding of anti-HER2/neu–sensitized (trastuzumab IgE) ADMC to SK-BR-3 cells induced significant cell death of the breast cancer cells as assessed by the uptake and visualization (red) of the PI in the SK-BR-3 cells but not the ADMC. Quantification of the PI signal in Figure 5B demonstrated significant breast cancer cell killing after 4 days (p = 0.0003). Sensitization of the ADMC with NS IgE did not result in significant SK-BR-3 cell death. Similarly, anti-HER2/neu IgE C6MH3-B1 sensitized ADMC induced significant (p = 0.032) SK-BR-3 cell death (data not shown). In addition, ADMC added to the SK-BR-3 over 4 days revealed significant (p = 0.003) breast cancer cell death, but not ADMC death (anti-tryptase labeled), as indicated by immunostaining of the SK-BR-3 with an Ab specific for the apoptotic enzyme caspase 3 (Figures 5C,D). Lastly, a significant (p = 0.0004) increase in caspase 3/7-positive breast cancer cells was confirmed at day 4 when anti-HER2/neu–sensitized ADMC were co-incubated (Figures 5E,F). The tumor cell specificity of the responses was verified as the NS isotype control IgE did not affect breast cancer cell viability. These experiments indicate ADMC binding to SK-BR-3 results in ADMC activation through FcεRI capable of inducing significant SK-BR-3 cell death.

As shown above, ADMC produce mediators that induce significant breast cancer cell death upon FcεRI cross-linking using a tumor-targed IgE. The ability of the mediators obtained from FcεRI-activated ADMC alone to induce SK-BR-3 cell death was then examined. As seen in Figures 6A,B, medium alone (not containing ADMC) from optimally activated ADMC incubated with SK-BR-3 cells induced significant (p = 0.009) SK-BR-3 cell death when incubated for 4 days. Further, when the media from optimally activated FcεRI ADMC were added to the SK-BR-3, there was a significant (p = 0.01) increase of apoptotic cells as evidenced by the increase in active caspase 3 (Figures 6C,D) indicating cell death of the breast cancer cells as in Figure 5. A significant (p = 0.0002) increase in activated caspase 3/7-positive breast cancer cells was confirmed at day 4 when SK-BR-3 cells were incubated with supernatants from ADMC activated through FcεRI (Figures 6E,F). Blocking TNF-α activity significantly prevented SK-BR-3 cell death (Figure 6G).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.