Tumors modulate fenestrated vascular beds and host endocrine status

Allograft and xenograft transplantation into a mouse host is frequently utilized to study cancer biology, tumor behavior, and response to treatment. Preclinical studies employing these models often focus solely upon the intra‐tumoral effects of a given treatment, without consideration of systemic toxicity or tumor–host interaction, nor whether this latter relationship could modulate the toxicologic response to therapy. Here it is demonstrated that the implantation and growth of a range of human‐ and mouse‐derived cell lines leads to structural vascular and, potentially, functional changes within peripheral endocrine tissues, a process that could conceivably ameliorate the severity of anti‐angiogenic‐induced fenestrated vessel attenuation. Observations suggest a multifactorial process, which may involve host‐ and tumor‐derived cytokines/growth factors, and the liberation of myeloid‐derived suppressor cells. Further investigation revealed a structurally comparable response to the administration of exogenous estrogen. These findings, in addition to providing insight into the development of clinical anti‐angiogenic “adaptation,” may be of significance within the “cancer‐cachexia” and cancer‐related anemia syndromes in man.


| INTRODUCTION
The past decades have witnessed huge advances in the understanding of cancer growth and progression, allowing the development of multiple targeting modalities in preclinical oncology research (Seebacher et al., 2019). Much of this work has utilized allograft and xenograft transplantation in mice. Although this has led to breakthroughs in new treatments, these efficacy and pharmacodynamic studies have until recently experienced limited utility in the field of toxicology research, that is, the safety testing of drugs within a diseased population (Kim & Sharpless, 2012;Morgan et al., 2017Morgan et al., , 2013. In the field of vascular endothelial growth factor (VEGF)-inhibiting and other anti-angiogenic therapies, preclinical animal models have been used to identify treatment-related microvascular lesions, which may contribute to a range of dose-limiting and potentially severe toxicities, clinically, for example, hypertension, proteinuria, hemorrhage/ thrombosis, gastro-intestinal perforation, and endocrine dysfunction (Brinda et al., 2016;Chen & Cleck, 2009;Hanna et al., 2020;Hayman et al., 2012;Izzedine et al., 2009;Kamba et al., 2006;Kamba & McDonald, 2007;Pena-Hernandez et al., 2019;Syrigos et al., 2011;Touyz et al., 2018;Yang, Zhang, et al., 2013;Zhang et al., 2016).
A complete understanding of the factors responsible for the de novo formation of tumor-associated blood vessels in cancer is currently lacking. Additionally, the question remains as to whether similar factors could impact the vasculature in host tissues, via comparable mechanisms, and whether or not this could also influence the vascular response during therapy (Grunewald et al., 2006;Lugano et al., 2020;Zuazo-Gaztelu & Casanovas, 2018).
In addition to providing insight into the pathogenesis of cancerendocrine comorbidity, an enhanced understanding of such processes could aid in the prediction of how a given vascular modulating agent (VMA)-treated cancer patient may respond systemically to angiogenic inhibition.
We have previously demonstrated that both the nude mouse and the C57Bl immunocompetent mouse respond similarly to toxicologic modification of systemic endocrine vasculature via the administration of VMAs (Hargreaves et al., 2017).
During these preliminary investigations, we determined that the transplantation and growth of a Calu-6 xenograft tumor, in the absence of any treatment, could potentially increase resident vascular density within host endocrine tissues ( Figure S1). The aim of this manuscript was to test a hypothesis that this observation within fenestrated endocrine vascular beds may occur in association with other tumor types, of variable lineage, and with differential expression of angiogenic signaling mediators.
Here we further demonstrate that the implantation and growth of a range of human-and mouse-derived xenografts/allografts leads to structural vascular and, potentially, functional changes within peripheral endocrine tissues. Observations suggest a multifactorial process, which may involve host-and tumor-derived cytokines/growth factors, and the liberation of myeloid-derived suppressor cells (MDSCs).

| Tumor-bearing animal models
In order to examine the systemic vascular and endocrine effects of tumor burden, a series of studies were conducted in nontumor-bearing and tumor-bearing, untreated, mice.
All mice were supplied by the Rodent Breeding Unit (AstraZeneca, Alderley Park, UK) and housed in negative pressure isolators with 12-h light/dark cycles and provided with sterilized food (supplied by Special Diet Services, Alderley Park, UK) and water ad libitum. The mice weighed approximately 25 g and were at least 8 weeks of age at study commencement. Animals were randomized into either control non-tumor-bearing or tumor-bearing groups (n = 6 in each) (Table 1) For BT474c studies, mice were supplemented with 0.36 mg/60 day release 17β estradiol pellets (Innovative Research of America, Florida, USA), 1 day prior to cell implantation. Two control groups, with and without estradiol supplementation, were included within this study cohort.
Following study completion, all mice were culled by terminal narcosis with 5:1 CO 2 /O 2 mixture.
All procedures were conducted in accordance with Home Office (UK) and local ethical review committee guidelines and complied with the Animals Scientific Procedures Act 1986.
2.2 | Tissue harvesting, processing, and vascular/ cell quantification Following the in life phase, selected endocrine organs (right adrenal gland, thyroid gland, and pancreas) were removed from all animals (the adrenal gland weighed). Tissues were fixed, embedded, and processed for CD31 immunohistochemical (IHC) staining and 3D fiberlength density analysis, using the Definiens XD image analysis platform (version 2.0.4 Definiens AG), with Tissue Studio and Developer XD, as described previously (Hargreaves et al., 2017). For CD11b and GR-1 (MDSC) expression, variations in slide preparation were as follows. For CD11b expression, slides were incubated overnight at 4 C with the primary antibody, rat anti-mouse CD11b (clone M1/70; BD Biosciences, New Jersey, USA), diluted at 1:40. For GR-1 expression, slides were incubated overnight at 5 C with the primary antibody, rat anti-mouse GR-1 (clone RB6-8C5, eBioscience, California, USA), diluted at 1:80. For both protocols, rat IgG2b was used to establish the appropriate isotype control. Discovery Red (#760-228; F. Hoffmann La-Roche, Basel, Switzerland) and Discovery Yellow (#760-239) chromogens were used to visualize the expression of CD11b and GR-1, respectively. Due to the inherent paucity of CD11b and GR-1-positive cells within the systemic endocrine tissues, a total T A B L E 1 Animal model groups incorporated to explore the effect of different tumor types upon peripheral host vascular bed density  LDS-751, to four wells, respectively, before analogous processing).

Analysis was performed using a Becton Dickinson FACSCanto II, with
FACSDiva software (BD Biosciences, New Jersey, USA), as per the manufacturer's instructions, generating scatter plot cytograms.

| Statistics
For statistical analysis, data were examined for normal distribution using the Shapiro-Wilk test. All comparisons containing normally distributed data groups were subject to unpaired t test analysis with Welch's correction (with assumed unequal variance). Comparisons of any groups found to have a non-Gaussian distribution were completed using Mann-Whitney non-parametric analysis. The significance level for all tests was 0.05.

| RESULTS
Tumor growth was well-tolerated in all animals over the duration of study, with no premature decedent animals, nor clinical signs indica-

| PC3 tumors
Automated analysis of 3D fiberlength density revealed tumor-related increases in the vascularity of thyroid interstitial tissue amongst PC3 tumor-bearing mice, when compared with the thyroid glands from non-tumor-bearing nu/nu animals. These animals also showed a decrease in TSH serum concentration. Multiplex serology also displayed a significant release of xenograft (human)-derived angiopoietin-2, G-CSF and IL-8, and FACS analysis of the bone marrow revealed reductions in erythrocyte and lymphocyte cell lineage numbers, with a concomitant increase in myeloid cells, leading to an increased myeloid:erythroid ratio, amongst PC3 prostate xenograft tumor-bearing animals ( Figure 3).

| 4T1 tumors
Automated analysis of 3D fiberlength density revealed increases in the vascularity of the adrenal gland and thyroid interstitial tissue, amongst 4T1 tumor-bearing mice, when compared with non-tumor-bearing BALB/c mice. There was also a decrease in serum concentrations of both ACTH and TSH amongst these mice. Multiplex serology revealed a significant increase in the serum concentrations of murine G-CSF, IL-6, PlGF-2, and prolactin.
As with the PC3 xenograft, FACS analysis of the bone marrow from the mice carrying the 4T1 syngeneic breast tumor showed an increase in the myeloid:erythroid ratio. However, the increase was even greater than that seen with the PC3 mice, due to a greater decrease in the immature nucleated red blood cell component (Figure 4). Preliminary investigations utilizing the Calu-6 xenograft tumor, in the development of structural biomarkers of vascular modulation, led to our hypothesis that additional "tumor-host interactions" may be responsible for structural and, potentially, functional changes within peripheral endocrine tissues, notably those with a fenestrated endothelial vascular plexus (Kamba et al., 2006;Kamba & McDonald, 2007).
Here it is demonstrated that the implantation and growth of a range of mouse-and human-derived allografts/xenografts in mice not only lead to structural vascular changes within the peripheral endocrine tissues but may also impart functional consequence.

F I G U R E 4
Plot showing that automated estimation of 3D fiberlength density detected significant increases in adrenal gland and thyroid interstitial vascularity amongst 4T1 tumor-bearing animals, when compared with non-tumor-bearing control mice (A,B). Pancreatic islet vascular densities were similar between the two groups. This was associated with increased serum concentrations of murine granulocyte colony stimulating factor (G-CSF), interleukin (IL)-6, PlGF-2, and prolactin (C-F), and with an apparent decrease in circulating adrenocorticotropic hormone (ACTH) and thyroid-stimulating hormone (TSH) (G). Fluorescence-activated cell sorting (FACS) analysis (H) revealed a significant shift in myeloid:erythroid ratio (MER) (increased) amongst 4T1 tumor-bearing mice. Dot plot showing total nucleated cell (TNC) and red blood cell (RBC) distribution from a non-tumor-bearing control BALB/c mouse (I) and a 4T1 tumor-bearing mouse (J). This demonstrates a sharp reduction in nucleated cells of erythroid lineage, with a concomitant increase and wider distribution of precursor/progenitor cells of myeloid lineage, in the tumor-bearing animal. Red = red blood cells, blue = nucleated erythroid cells, pink = lymphoid cells, green = myeloid cells. Median ± 10th-90th percentile for six animals per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Histogenetically diverse subtypes of tumor appeared stratified in their ability to influence peripheral endocrine vascular bed density.
Specifically, CT26 (syngeneic colorectal) tumors were not associated with structural vascular alteration amongst peripheral endocrine tissues, nor with serum changes in ACTH/TSH. These findings were associated with a relatively "weak" release of the murine pro-angiogenic cytokines/growth factors analyzed and the absence of a notable myeloid reaction. The 4T1 (syngeneic breast) tumor growth was associated with enhanced adrenal cortical and thyroid interstitial vascularity, and decreased ACTH and TSH serum concentrations. The increased serum concentrations of murine G-CSF, IL-6, PlGF-2, and prolactin were consistent with the production of a series of growth factors/hormones, with varying angiogenic potency (Corbacho et al., 2002;De Falco, 2012;Gopinathan et al., 2015;Nagasaki et al., 2014;Nejabati et al., 2017;Reuwer et al., 2012). However, as this was a syngeneic model, it was unclear, within the context of this study, as to the exact source of these proteins, that is, whether liberated from host tissues . Note that, in normal tissues, these cells were found in exceptionally low numbers, with a significant elevation within the tissues of 4T1 tumor-bearing animals. High-power CD11b (C,D) and GR-1 (E,F) stained sections of adrenal gland from non-tumor-bearing (C,E) and 4T1 tumor-bearing (D,F) mice (original objective magnification 10Â, scale bar = 100 μm). Note the presence of intra-sinusoidal positive cells within the adrenal cortex of the tumor-bearing example (arrows). Median ± 10th-90th percentile for six animals per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 et al., 1994;Yan et al., 2013;Yang, Meyer, & Friedl, 2013;Yonekura et al., 1999;Zhao et al., 2008).
The 4T1 implantation and growth also induced a marked shift in myeloid:erythroid ratio within the bone marrow, in part due to a dramatic decrease in nucleated erythroid cell production. Interestingly, the role of IL-6 in the suppression of erythrogenesis is well documented; the serum concentrations of this cytokine were significantly elevated within the 4T1 study, recapitulating the known role of this F I G U R E 6 Plot showing that automated estimation of 3D fiberlength density detected significant increases in adrenal gland (A), pancreatic islet (B), and thyroid interstitial (C) vascularity amongst both non-tumor-bearing and BT474c tumor-bearing mice exposed to exogenous estrogen administration, when compared with non-tumor-bearing, non-estrogen-supplemented, control mice. A significant release of host-/xenograftderived granulocyte colony stimulating factor (G-CSF)/interleukin (IL)-8 (D,E) was observed, along with estrogen-associated increases in leptin (F; significance in non-tumor-bearing animals only), and prolactin (G). Serological probing for adrenocorticotropic hormone (ACTH) and thyroidstimulating hormone (TSH) (H) displayed a relative decrease in both circulatory ACTH and TSH amongst BT474c tumor-bearing, estrogensupplemented, animals. In addition, there was an apparent decrease in circulatory TSH amongst non-tumor-bearing nude mice subject to supraphysiological estrogen exposure alone. There was a significant reduction in all bone marrow cell lineages amongst estrogen-supplemented animals, as demonstrated by fluorescence-activated cell sorting (FACS) analysis (I). There was therefore no resultant trend in myeloid:erythroid ratio amongst estrogen-supplemented animals, with or without a tumor burden. These changes were additionally unassociated with increased CD202b-positive cell production at this time-point (data not shown). Median ± 10th-90th percentile for six animals per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 protein in the pathogenesis of anemia (Ershler, 2003;Macciò et al., 2005;Madeddu et al., 2018). IHC for CD11b and GR-1 indicated a significant influx of cells with morphologic and staining characteristics consistent with MDSCs into the peripheral endocrine sinusoidal vasculature of tumor-bearing mice. The contribution of bone marrow-derived cells, to the promotion of neoplastic angiogenesis, is well-established (Coussens & Werb, 2002;Gabrilovich, 2017;Hsu et al., 2019;Murdoch et al., 2008;Varner & Schmid, 2010;Wesolowski et al., 2013). Interestingly, the liberation of MDSCs has also been documented in myriad other non-neoplastic disease contexts, such as trauma, endotoxemia, radiation, thermal injury, myocardial infarction, and hepatic fibrosis (Linz et al., 2013;Song et al., 2018;Suh et al., 2012;Thanasegaran et al., 2015;Van Rompaey & Le Moine, 2011;Wu et al., 2014;Yao et al., 2015). Although the cellular source of murine G-CSF was not designated within the context of these studies, previous work has demonstrated an enhanced liberation of CD11b + ve/GR-1 + ve MDSCs, via the intra-tumoral production of G-CSF, in the development of tumor refractoriness to anti-VEGF therapy in mice, and highlighted this cytokine as a potential therapeutic target (Li et al., 2016;Morales et al., 2010;Shojaei & Ferrara, 2008;Shojaei et al., 2009) poorly understood, effect of supra-physiological estrogen administration (Abid et al., 2017;Farris & Benjamin, 1993;Gaunt & Pierce, 1985;Jilka et al., 1995;Macneil et al., 2011;Sontas et al., 2009). There is a well-established association between ER activation and the promotion of angiogenesis, and/or enhanced blood flow, within many physiological and pathological disease states,  Losordo & Isner, 2001;Seo et al., 2004;Sohrabji, 2015).
Importantly, blocking the ER cascade has also recently emerged as a modality by which to potentially abrogate the development of antiangiogenic resistance (Gu et al., 2020). Within this study, we were unable to demonstrate the involvement of CD202b + ve EPCs within these observations. However, this cell type may only be transiently released with estrogenic stimulation; such events may have preceded the time-point at which bone marrow aspirates were harvested (Akwii et al., 2019;Masuda et al., 2007;Rajoria et al., 2011;Suriano et al., 2008). Further investigation would be needed to delineate the functional and structural sequelae involved in this observation of steroid-mediated sinusoidal "hypervascularity." When linked to an apparent effect upon host endocrine status, the presence of a tumor burden appeared to be associated with a variable but relative decrease in host terminal ACTH and TSH serum con- Interference in ACTH synthesis and release and/or changes in cortisol/corticosterone metabolism have been suggested as contributory, with exaggerated adrenal blood flow, a finding consistent with the structural vascular findings presented here, also reported (Boonen et al., 2013;Boonen et al., 2015;Kanczkowski et al., 2015;Kanczkowski et al., 2017;Lang et al., 1984;Peeters et al., 2015). Similarly, depressed thyroid gland function has also been previously documented in experimental models of cancer anorexia-cachexia.
However, as in the clinical scenario, confounding effects of decreased food intake and/or weight loss have been variably linked to the observed decreases in thyroid action (Costelli & Baccino, 2000;Persson et al., 1985;Svaninger et al., 1986). Weight loss and inappetence were not recorded during the studies outlined here. Within this study, both ACTH and TSH serum concentrations were measured at the terminal sacrifice time-point. Although a direct and potentially vascularmediated action upon the adrenal and thyroid glands is possible as a novel mechanism of cancer-induced host endocrine modulation, further work would be required to delineate these findings in greater detail. This would ideally incorporate the sequential monitoring of both pituitary stimulation (ACTH/TSH), and end-organ function (corticosteroid/thyroxine production), in addition to further exploration around other confounding factors that may influence these signaling axes (De Vries et al., 2016;Gong et al., 2015;Louis et al., 2017;Romanò et al., 2020;Wondisford, 2018).

| CONCLUSION
Here we demonstrate that the implantation and growth of a range of human-and mouse-derived tumors leads to structural vascular and, potentially, functional signaling changes within host mouse endocrine tissues. Tumor-and host-derived cytokines/growth factors, and MDSC liberation, are implicated in this phenomenon, and these observations may be of significance within the development of clinical anti-angiogenic "adaptation." Further study will expand knowledge of the potential for cancer and supra-physiological steroid administration to directly influence the vascular structure and function of systemic endocrine tissue, elucidate the role of these findings within the multifaceted "cancercachexia" and cancer-related anemia syndromes, and clarify the clinical significance regarding on-target endocrine vascular toxicity, and apparent modification in the presence of varying tumor types, in man.

CONFLICT OF INTEREST
The authors have no conflicts of interest to declare.

FUNDING INFORMATION
This manuscript was co-sponsored by AstraZeneca PLC, PathCelerate Ltd., and the University of Surrey. There are no grants/award numbers to declare.

SUPPLEMENTARY FIGURE
During preliminary method development investigations (Hargreaves et al., 2017), automated analysis of 3D fiberlength density revealed an increase in adrenal cortical and pancreatic islet vascularity amongst Calu-6 tumor-bearing animals, when compared with the cohort of non-tumor-bearing animals. There was no significant increase in thyroid interstitial vascularity ( Figure S1).