Chronic Stress Promotes an Immunologic Inflammatory State and Head and Neck Cancer Growth in a Humanized Murine Model

Abstract

Background:

Despite the importance of immune response and environmental stress on head and neck cancer (HNC) outcomes, no current pre-clinical stress model includes a humanized immune system.

Methods:

We investigated the effects of chronic stress induced by social isolation on tumor growth and human immune response in subcutaneous HNC tumors grown in NSG-SGM3 mice engrafted with a human immune system.

Results:

Tumor growth (p<0.0001) and lung metastases (p = 0.035) were increased in socially isolated versus control animals. Chronic stress increased intra-tumoral CD4⁺ T cell infiltrate (p = 0.005), plasma SDF-1 (p<0.0001) expression, and led to tumor cell dedifferentiation towards a cancer stem cell phenotype (CD44⁺/ALDH^high, p=0.025).

Conclusions:

Chronic stress induced immunophenotypic changes, increased tumor growth, and metastasis in HNC in a murine model with a humanized immune system. This model system may provide further insight into the immunologic and oncologic impact of chronic stress on patients with HNC.

Introduction

The annual rate of head and neck cancer (HNC) in the United States exceeds 66,000 cases resulting in over 14,000 associated deaths per year¹. Surgery and/or radiotherapy remain principal management strategies for HNC^2–4. However, despite optimal management, long-term survival rates remain low and prognosis is particularly poor in certain sociodemographic groups. Several social determinants of health have been independently associated with worse cancer control in HNC^5–9. In particular, patients with low socioeconomic status (SES) present with more aggressive disease than patients with high SES. This includes increased nodal involvement, more distant metastasis, higher tumor grade, and worse locoregional control^8,9. While these findings may be partly explained by clinical factors such as poor quality care or decreased access to care, the disparity in treatment outcomes based on SES persists even when adjusting for relevant covariates including demographic, tumor, and treatment factors^10,11. This suggests that low SES, for undefined reasons, is independently associated with an inherently aggressive HNC pathobiology leading to worse oncologic outcomes.

This link between SES and cancer biology in HNC may be mediated by the impact of chronic psychosocial stress in low-SES patients¹². Under a variety of adverse life circumstances, including low SES, social deprivation, and social isolation, patients’ circulating blood leukocytes demonstrate a conserved transcriptional response to adversity (CTRA) involving upregulation of proinflammatory markers^13–17. A link between this elevated inflammatory state in low-SES patients and worse oncologic outcomes has been demonstrated in several other malignancies, including hematologic and breast cancers^18–20. These findings have been further replicated in pre-clinical studies. In a variety of tumor types, social adversity in animal models has been associated with more aggressive cancer growth rates and treatment resistance^21–23. This increased growth has been attributed largely to beta-adrenergic pathways.

However, prior pre-clinical models establishing this relationship have used either immunocompromised mice or syngeneic models^24–29. Both of these models have important limitations. Immunodeficient mice or those with an intact murine immune system do not recapitulate the human tumor microenvironment and results in these models do not translate well to human clinical trials^30,31. This is evidenced by the large number of pre-clinical studies using chronic stress models of HNC which have identified beta-adrenergic stimuli as the primary driver of stress-induced cancer growth^29,32–36. Despite such overwhelming pre-clinical evidence for the benefit of beta-blocking agents, clinical use of beta-blockers has not demonstrated any substantial anti-neoplastic activity against HNCs and has, conversely, been associated with decreased survival^37–39.

This suggests that beta-adrenergic stimuli alone are not the primary driver of tumor growth and metastasis in the setting of chronic stress. We hypothesized that additional factors in the tumor immune microenvironment contribute to the increased tumor growth in HNC related to chronic stress stimuli. Therefore, to more reliably investigate how chronic stress affects HNC growth in humans, we developed a novel pre-clinical model of chronic stress using a human HNC xenograft in a murine model with a humanized immune system.

Materials and Methods

Patient Peripheral Blood Mononuclear Cells

To explore the association between chronic socioeconomic stress in HNC and systemic inflammation in the clinical context, cryopreserved patient peripheral blood samples were obtained from patients which consented to cell banking in the Medical College of Wisconsin Tissue Bank using an IRB-approved protocol. Mononuclear cells were purified by Ficoll-paque density gradient separation.

Animals

All experimental procedures were carried out according to the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee (IACUC) of the Medical College of Wisconsin. A breeding colony of NSG-SGM3 mice was established from breeder pairs purchased from Jackson Laboratory (strain: 013062). Equal numbers of male and female mice aged 4–6 weeks were used for each study arm.

Humanization with cord blood CD34+ cells

For humanization, mice received total body irradiation (TBI) to a dose of 2 Gy. Irradiations were performed between 8–10 AM on unanesthetized mice using a SmART Precision X-ray instrument (225 kVp; 10 mA; half value layer (HVL) 0.89 mm Cu) with a dose rate of 107.7 cGy/min. At 24 hours post-irradiation, animals were transplanted via tail vein with a dose of 50,000 CD34+ cells from mixed donors purchased from StemCell technologies (catalog number: 70008). At four weeks post-CD34+ transplant, human CD45+ chimerism was determined by FACS analysis of peripheral blood. The following antibody panel was used: APC anti-human CD45 (Biolegend, 3368512), Pacific Blue Anti-human CD3 (Biolegend, 300330), FITC anti-mouse CD45 (Biolegend, 103108), PE anti-human CD13 (Biolegend, 301704), 7-AAD (Biolegend, 420404), APC-CY7 anti-human CD19 (Biolegend, 302218).

Inducement of Chronic Stress by Social Isolation

Following determination of human peripheral blood CD45+ engraftment, animals were assigned to either normal housing or social isolation treatment arms. Groups were assigned such that human engraftment was similar between groups and equal numbers of male and female mice were included in each arm (5 male and 5 female). For the normal housing controls, animals were maintained with their original littermates in groups of 2–3 animals per cage. For social isolation, animals were separated from their original cage mates and singly housed for 4 weeks prior to implantation of subcutaneous tumors.

Measurement of Inflammatory Gene Expression

To investigate whether the systemic effects of chronic environmental stress were analogous in experimental and clinical contexts, mononuclear cells from either patient peripheral blood samples or murine blood were isolated by red blood cell lysis. RNA was extracted using the Qiagen RNeasy micro kit (Qiagen, Germany). RNA was reverse transcribed into cDNA using iScript cDNA synthesis kit. Real time PCR analysis was performed using Taqman Gene Expression assays (Life Technologies, Carlsbad, CA) on a Biorad CFX96 PCR Machine. Data were normalized to GAPDH and littermate controls using the delta-CT method. Fold change relative to control group was determined for each gene.

Implantation of Subcutaneous FaDu tumors

The hind flank was shaved with an electric razor. A 200 uL suspension of 2×10⁶ FaDu cells in Matrigel were implanted in the hind flank using an insulin syringe.

Histological Analysis

Hematoxylin and Eosin (H&E) and Ki-67 staining was performed at the Wisconsin Children’s Hospital Pathology core on formalin-fixed paraffin embedded 5 μm sections from 5 animals per treatment group. Enumeration of total lung metastases in each H&E stained lung lobe section was performed on an EVOS M5000 fluorescence imaging system at 4X. Quantification of percent Ki-67 cells (DAB positive cells) was performed using the Image J Colour Deconvolution plugin⁴⁰. For each tumor or lung, the %Ki-67 is reported as the average of five 10X fields per tumor or lung. Representative images were captured at 10X and 20X magnification.

Determination of Tumor Volume

Tumor growth was monitored by external digital caliper measurements, twice weekly from day 0 to day 28. Tumor volume was calculated as π(lxw²)/6. Where the l is the length of the tumor long axis and w is the length of the tumor short axis⁴¹.

Tumor Processing

At the 28 day endpoint, tumors were carefully dissected and cut in half lengthwise. One half of the tumor was preserved in 10% buffered formalin for subsequent histology analysis. The remaining tumor was diced into 3 mm segments using a razor blade. The tumor was then digested according to the Miltenyi Tumor dissociation kit (130-095-9291) protocol in a Miltenyi gentleMACS C tube using the Miltenyi gentleMACS tissue dissociator. Cellular debris was removed from the dissociated cell suspension using Debris Removal Solution (Miltenyi 130-109-398). Live cells were enumerated using trypan blue dye on Countess Automated Cell Counter Hemocytometer C10227 (Invitrogen). Human CD44 tumor cells were isolated by magnetic column purification (Miltenyi, 130-095-194) per the manufacturer’s protocol.

FACS Analysis

One million cells/sample were used for the surface staining with conjugated monoclonal antibodies. 7-AAD (420404, Bio Legend) or Viobility 405/520 fixable dye (130-109-814, Miltenyi) was used to eliminate dead cells. Single color tubes were used to set up compensation matrix, and a Fluorescence Minus One (FMO) control tube was included to ensure specific staining. For immunophenotyping of digested tumors, the following antibodies were used: APC-Vio770 anti-human CD45 (MIltenyi, 130-110-635), FITC anti-human CD4 (Miltenyi, 130-114-531), PerCP-Vio700 anti-human CD3 (Miltenyi, 130-113-141), VioBlue anti-human CD8 (MIltenyi 130-110-683). The antibody staining was done at 4°C for 30 min. Cells were washed (1xPBS with 10% PBS) and suspended in 0.2 mL staining buffer (1XPBS with 2% FBS). Human ALDH activity staining within column purified CD44 cells was performed used the ALDEFLUOR kit (StemCell Technologies, 1700) per the manufacturer’s instructions. The samples were then acquired on a MACSQuant 10 Analyzer Flow Cytometer (Miltenyi). Data was analyzed using FlowJo software version 10.0 (BD Life Sciences). To verify gating and purity, all populations were routinely backgated.

Statistical Analysis

Values are reported as mean ± SEM, unless stated otherwise. All comparisons were performed using GraphPad Prism 9.0. All data were checked for normal distribution and similar variance between groups. Data are derived from multiple independent experiments from distinct mice. All animal studies were performed using sex and age-matched animals. Animal studies were performed without blinding of the investigator and no animals were excluded from the analysis as outliers.

Results

Profile of Peripheral Blood Inflammatory Gene Expression in Head and Neck Cancer Patients

To determine if low-SES in HNC patients correlates with an increase in inflammatory gene expression in circulating peripheral blood mononuclear cells, we obtained cryopreserved HNC patient blood samples from the MCW Tumor Bank. Blood samples were taken at the time of surgical resection. Patients were stratified by household income as either low-SES (<$45,000) or high-SES (>$45,000). Our analysis included three samples from low-SES patients (average annual income $39,000 ± $2,000) and three samples from high-SES patients (average annual income $91,000 ± $5,000). We measured mRNA transcript levels of five pro-inflammatory genes: TNF-α, IL-6, IL-8, IL-1A, and IL-1B. We observed marked and significant fold increases in mRNA expression in TNF-α, IL-6, IL-8, and IL-1A in the three low-SES patients compared to the high-SES patients (Fig. 1A).

Figure 1.

Chronic stress in humanized mice induces a similar pattern of immune cell pro-inflammatory gene expression as observed in low-SES (low-socioeconomic status) patients. A, Quantitative real-time PCR analysis of human low-SES and high-SES patients. Data are represented as fold-change relative to high-SES for each gene. B, Schematic illustrating protocol for NSG-SGM3 mouse humanization (top). Quantification of percent human CD45⁺ cells present in the murine peripheral blood (bottom left). Quantification of percent human CD3⁺, CD19⁺, CD13⁺ cells within the human CD45⁺ fraction (bottom right). Data are presented as mean ± SD. (N=9 mice per group). C, Quantitative real-time PCR analysis of peripheral blood mononuclear cells from chronically stressed and control mice. Data are represented as fold-change relative to control for each gene (N=4 mice per group). P-values for (A) and (C) are determined by individual t-tests with a Holm-Sidak multiple testing correction.

Development of Humanized Mouse Model of Chronic Stress

To assess the role of a chronically stressed human immune system in regulating HNC tumor growth in vivo, we developed the following preclinical mouse model. Sublethally irradiated (2 Gy) immunodeficient NSG-SGM3 mice were transplanted intravenously with 50,000 human cord blood CD34⁺ cells (Fig 1B, top schematic). At four weeks post-transplant, the percentage of human CD45⁺ blood cells in the peripheral blood (PB) were quantified by FACS analysis (Fig 1B, lower left). Human CD45⁺ PB chimerism was 23 ± 2%. Within the engrafted human CD45⁺ cell population, mature CD3 T-lymphocytes, CD19 B-lymphocytes, and CD13 myeloid cells were present. Twenty animals were divided evenly into control (n=10) or chronic stress (n=10) groups such that there was equal sex and human PB chimerism distribution. Animals assigned to the control stress protocol were removed from their original housing and transferred to single housing for four weeks. Control animals remained in social housing with their original littermates. After four weeks of social isolation, the peripheral blood mononuclear cell transcript expression of human TNF-α, IL-6, IL-8, IL-1A, and IL-1B was measured. Chronic stress increased transcript levels of all inflammatory factors, with significant increases observed in IL-6, IL-8, and IL-1A compared to control animals (Fig. 1C).

FaDu Tumor Growth in the Humanized Mouse Model of Chronic Stress

We then assessed whether subcutaneous tumor growth of the human HNC FaDu cell line is affected by the stressed human immune microenvironment. Both humanized stressed and control animals were additionally compared to non-humanized immunodeficient control animals. Here, 2×10⁶ FaDu cells suspended in Matrigel were implanted in the hind flank. One humanized animal in each group (control and stress conditions) died of an unrelated disease process prior to tumor growth and were excluded from data analysis. Tumor growth was assessed until the 28 day endpoint was reached. The rate of tumor growth and tumor size at the 28-day endpoint was largest in non-humanized immunodeficient FaDu control animals compared to either humanized treatment group (Figs 2A–C). An exponential growth curve was fit to each data set to determine tumor doubling time (Fig 2A, dashed lines). Tumor doubling time in the non-humanized FaDU group was 5.2 days (95% CI: 4.4–6.0). In humanized animals, doubling time was increased to 6.2 days (95% CI: 5.3–7.1) and 5.9 days (95% CI: 5.0–6.9) in humanized Fa-Du mice (Hu-FaDu) and chronically stressed Hu-FaDu mice, respectively. In humanized mice, the tumor volume at endpoint was significantly larger in the stressed animals when compared with humanized Fa-Du control animals (Figs 2B, C). Commensurate with the increase in tumor size, the percentage of tumor Ki-67+ proliferative cells was significantly increased in chronically stressed mice compared to unstressed Hu-FaDu mice (Fig 2D). In the humanized model, we observed histological evidence of tumor vascularization and metastasis to the lung in both the control and socially isolated groups (Fig 2E–F). Chronic stress significantly increased the number of metastatic lesions per lung lobe compared to humanized control animals (Fig 2E). Additionally, the percentage of Ki-67+ proliferative cells in the lungs of chronically stressed mice was significantly higher than unstressed mice (Fig 2F).

Figure 2.

Chronic stress in humanized mice accelerates tumor growth and promotes metastasis. A, FaDu subcutaneous tumor volume growth curves for non-humanized (N=5), humanized control (N=9), and humanized with chronic stress (N=9). Solid data points represent mean +/−SEM. Dashed lines represent exponential best fit curve. B, Representative gross tumor pictures at day 28 endpoint. C, Tumor volumes at day 28 endpoint (bar indicates median, p-values determined by one-way ANOVA with Tukey’s multiple comparison test). D, Left, representative Ki-67-stained sections of the primary tumors (scale bar = 100 μm). Right, quantification of percent Ki-67+ tumor cells (N=5 tumors per treatment). E, Metastatic lung lesions per lobe in humanized control (N=9) and chronically stressed (N=9) mice (bar indicates median, p-value based on non-parametric Mann-Whitney test). F, Left, representative metastatic lung lesions in H&E (left) and Ki-67 (right) images in each treatment group (scale bars = 100 μm). Right, quantification of percent Ki-67+ tumor cells (p-value based on non-parametric ANOVA with Dunnett’s multiple comparison test).

Tumor Immune Cell Microenvironment

At the day 28 endpoint, humanized tumors were analyzed by FACS for human immune cell infiltration. Chronic stress significantly increased both the percentage and total number of human CD45⁺ cells within the tumor microenvironment compared to non-stressed humanized controls (Fig 3A, B). Human CD3+ T-lymphocytes were the largest fraction of CD45+ cells within both groups (Fig 3C). Within the CD3+ population, chronic stress significantly increased the percentage of CD4+ cells (Fig 3D) and the ratio of CD4+/CD8+ cells (Fig 3E) in the tumor. Finally, the effect of chronic stress on cancer stem cell activity was assessed by aldehyde dehydrogenase (ALDH) activity within cells expressing the putative cancer stem cell marker CD44. Here, chronic stress significantly induced increased ALDH activity within the CD44+ population (Fig 3F).

Figure 3.

Chronic stress alters tumor immune microenvironment. (A) Left, representative FACS histogram of tumor infiltrating human CD45 cells in humanized treatment groups. Quantification of percentage human CD45 in humanized control and humanized with chronic stress (N=9 per group; p-value determined by 2-tailed Student’s t-test). (B) Total human CD45 tumor infiltrating lymphocytes (N=9 per group; p-value determined by 2-tailed Student’s t-test). (C) Analysis of percent CD3, CD19, and CD13 cells within the human CD45+ fraction (N= 5 per group). (D) Left, representative FACS plots of human CD4 and CD8 cells within the CD3 population (N= 9 per group; p-value determined by 2-tailed Student’s t-test). Right, quantification of percentage CD4+ cells within the CD3 population. (E) Ratio of CD4/CD8 TIL (N= 9 per group; p-value determined by 2-tailed Student’s t-test). (F) Percentage ALDH+ within column purified CD44+ cells (N=5 per treatment group; p-value determined by 2-tailed Student’s t-test). (G) Peripheral blood human cytokine analysis. Cytokines with a 1.5-fold or greater difference between treatment groups are plotted. P-value determined by one-way ANOVA with Sidak’s multiple comparison test.

Plasma Human Cytokine Analysis

At the 28-day endpoint peripheral blood plasma was screened for a panel of 71 human cytokines (human cytokine 71-plex, Eve Technologies) to screen for potential pathways differentially regulated by chronic stress in this model. Of this panel, 40 human cytokines were detectable in the mouse plasma. The 14 cytokines with a fold-change of +/−1.5 or greater are shown in (Fig 3G). Of these cytokines, the chemokine SDF-1 (CXCL12) is highly and significantly increased in chronic stress mice relative to unstressed controls (p<0.0001).

Discussion

Chronic social isolation stress in this study led to increased growth and metastasis of HNC xenografts. While such increased tumor growth and metastasis is consistent with prior reports using immunodeficient or syngeneic models, to our knowledge this is the first report to demonstrate increased HNC growth and metastasis in a chronic stress model with a humanized immune system. This is significant because it provides a successful model system to study the effects of chronic stress on the tumor immune microenvironment in a context that more closely resembles the human clinical condition than any previous models. This is imperative given the discordance between prior pre-clinical work demonstrating an overwhelming association of beta-adrenergic stimuli with tumor growth in chronic stress modeling and the absence of clinical findings to support any benefit of beta-blocking agents in HNC^37–39. The humanized model system in this study demonstrated the presence of both human myeloid and lymphoid lineages as well as expression of at least 40 human cytokines. These data indicate that our model is likely to well-represent the human condition and better predict human clinical responses as compared to both immunodeficient and syngeneic models.

Our study findings also suggest that there is an important immunoregulatory effect of chronic stress leading to HNC proliferation. There was a significant increase in overall human immune infiltrate in the chronic stress group as compared with the control condition, despite increased tumor growth in stressed animals. This may be related to greater influx of regulatory components, such as myeloid-derived suppressor cells or regulatory T cells, or less functionally effective tumor-infiltrating lymphocytes. While the precise mechanism will require further investigation in future study, our results identified a significant pro-inflammatory change in peripheral blood mononuclear cells in stressed animals, suggesting a phenotypic change in circulating immune cells in response to chronic stress. Importantly, when this phenotype was assessed in HNC patients in this study stratified by household income as a metric for SES, a similar pro-inflammatory change in peripheral blood mononuclear cells was seen in low-SES patients. This suggests that chronic socioeconomic stress may induce a similar pro-inflammatory immune state as our chronic stress model system.

In addition to immunoregulatory alterations in the human immune response, this study also identified important changes in the tumor cells in response to chronic stress. In particular, tumors from stressed animals were found to develop a significantly greater proportion of cells with a cancer stem cell phenotype, as measured by the expression of CD44⁺/ALDH^{high 42–44}. Increased activity of this tumor cell subpopulation has been associated with increased aggressiveness in HNC including increased lymph node metastasis and worse disease-free survival^45,46. Thus, one mechanism by which chronic stress may increase tumor proliferation and invasion is through induction of stem-like pathways. It has recently been demonstrated in an immunodeficient mouse model of breast cancer that chronic stress leads to the induction of cancer stem cells via adrenergic pathways, inducing increased tumor growth⁴⁷. However, once again, immunodeficient modeling does not recapitulate the human tumor immune microenvironment and the induction of stem-like pathways may also be mediated by pro-inflammatory or regulatory immune cytokines.

We identified significant upregulation of circulating levels of SDF-1 in the chronic stress group when compared to the control condition. Upregulation of SDF-1 in response to chronic stress has not before been reported but may be related to increased tumor growth through several mechanisms. SDF-1 (also known as CXCL12) is a chemokine which primarily binds CXCR4 and is involved in lymphocyte migration and trafficking in both physiologic and pathologic states, including inflammatory disorders, tissue injury, and cancer⁴⁸. The binding of SDF-1 to CXCR4 may induce multiple intracellular pathways which lead to migration, proliferation, and cell survival. Further, CXCR4 expression has also been identified in stem cells in multiple organ systems and overexpression has been reported in several malignancies^49,50. Taken together, we hypothesize that SDF-1 secretion from pro-inflammatory immune cells in the chronic stress condition may lead to induction of stem-like properties and associated increased proliferation in HNC cells.

There are several limitations of this study and this humanized model system. First, these results remain correlational rather causational. While increased social isolation stress in our model does lead to increased tumor growth and metastasis with our results implicating a regulatory and inflammatory role of the immune response, the precise mechanism remains uncertain. Additionally, this study was performed with a single HNC tumor cell line and requires validation in additional cell lines as well as patient-derived xenografts. Finally, in this humanized murine model, the engrafted immune system is allogeneic in relation to the engrafted tumor cell line, leading to HLA-mismatch between tumor and human immune cells. While the true effects of this are unknown, prior studies have demonstrated that such HLA mismatch does not significantly hamper xenograft growth, and since our models are created with fresh mixed donor CD34+, HLA-mismatch should be equivalent between replicates and cell lines⁵¹.

Conclusions

We have successfully developed a novel humanized murine model of chronic stress for HNC. We have demonstrated increase in tumor growth and metastasis in response to chronic stress with significant intra-tumoral and circulating immunophenotypic changes. We have further identified tumor dedifferentiation in response to chronic stress, as evidenced by an increase in tumor cells exhibiting a cancer stem cell phenotype. This model system may provide further insight into the immunologic impact of chronic stress on worse outcomes in patients with HNC who experience chronic stress conditions, including social adversity and low SES.