Expression of Myoglobin in Normal and Cancer Brain Tissues: Correlation With Hypoxia Markers

Background Myoglobin (MB) is increasingly recognized as a key player in cancer growth and metastasis. Low oxygen tensions, commonly associated with highly aggressive and recurrent cancers, have been shown to regulate its expression in several cancers such as lung, neck, prostate and breast cancer. However, it is not yet known whether it contributes to the growth and spread of brain cancers especially Glioblastoma multiforme (GBM). Methods Here we investigate the expression of MB, and its correlation with the hypoxia markers carbonic anhydrase IX (CAIX) and lactate dehydrogenase A (LDHA), in human tissue microarrays of multiple organ tumors, brain tumors, and GBM tumors, and their respective cancer-adjacent normal tissues. Correlation between MB protein expression and tumor grade was also assessed. Results We show that MB protein is expressed in a wide variety of cancers, benign tumors, cancer-adjacent normal tissues, hyperplastic tissue samples and normal brain tissue, and low oxygen tensions modulate MB protein expression in different brain cancers, including GBM. Enhanced nuclear LDHA immune-reactivity in GBM was also observed. Finally, we report for the first time a positive correlation between MB expression and brain tumor grade. Conclusion Our data suggest that hypoxia regulate MB expression in different brain cancers (including GBM) and that its expression is associated with a more aggressive phenotype as indicated by the positive correlation with the brain tumor grade. Additionally, a role for nuclear LDHA in promoting aggressive tumor phenotype is also suggested based on enhanced nuclear expression which was observed only in GBM.


INTRODUCTION
Glioblastoma multiforme (GBM) is the most common primary brain cancer and is associated with poor prognosis (1). The recently reported 2-and 5-year survival rates of GBM patients are 26 -33% and 4 -5%, respectively (2). Pathognomonic features of GBM include the presence of microvascular proliferation and necrosis (3), both indicative of hypoxia. Since hypoxic cancer cells have a more aggressive phenotype, understanding the mechanisms responsible for cancer cell survival under hypoxic conditions could help identify factors responsible for GBM aggressiveness and recurrence (4,5). Earlier reports show that GBM biopsies and xenograft derived cell lines display regional variation in oxygen consumption (6)(7)(8). Using cell lines from hypoxia-sensitive and hypoxia-tolerant GBM cells significant differences in their response to hypoxia were observed. The hypoxia-tolerant cell lines (M006x and M059K) were capable of reducing their oxygen consumption rates and maintaining their clonogenic potential after 4 days of hypoxia (0.6% oxygen) while the hypoxia-sensitive cells (M010b) failed to reduce their oxygen consumption rates and to maintain their clonogenic potential (6). These observations potentially explain GBM recurrence and ability to adapt to hypoxic tumor microenvironment, yet, the underlying hypoxia-adaptation mechanisms are still not clear.
The monomeric iron-and oxygen-binding protein, myoglobin (MB), was the first protein with a threedimensional molecular structure to be reported (9,10). MB has long been known to help in oxygen diffusion and storage and in both nitrite (NO2-) nitric oxide (NO) metabolisms in skeletal muscle and heart (11)(12)(13). Newer roles of MB are continuously being identified. For example, interaction with and possible aid in the intracellular transport of fatty acids as well as a potential role in cancer growth are being investigated (14).
MB expression has been reported in a wide variety of cancers such as human medullomyoblastoma and other non-myogenic epithelial cancers (15,16). For example, its mRNA and protein were detected in a subset of breast cancer cell lines and is upregulated in breast tumors compared to matched normal tissues (17,18). Similarly, normal epithelial cells of the colon, ovary and lung were negative for MB while a subset of their respective tumors showed variable levels of MB expression (17). MB is also expressed in renal cell carcinoma and is up-regulated in prostate cancers compared to normal adjacent tissues (19,20). Collectively these observations support a potential noncanonical role for MB in cancer growth.
In cancer, MB transcripts and/or protein are shown to be induced by different stimuli involved in cancer progression. For example, MB protein was induced by hypoxia, oxidative stress and exposure to epidermal growth factor in human MCF-7 breast cancer cell line, where oxidative stress was found to be the ultimate signal that triggers MB induction (17). In addition, MB expression significantly correlated with markers of hypoxia in a subset of invasive breast cancers (HIF-2a and CAIX) and in prostate cancer (CAIX) (18,20). In line with these findings, MB expression in clear cell renal carcinoma correlated negatively with capillary density and positively with tumor grading with MB transcript and protein levels being induced under hypoxia in different renal carcinoma cell lines (19). It is worth mentioning that hypoxia-mediated MB up-regulation isn't limited to cancer cells, in fact normal liver, brain, and gill of hypoxia-tolerant carp also up-regulate non-muscle type MB following five days of hypoxia (21).
We have previously reported that neuroglobin (Ngb) and cytoglobin (Cygb) are expressed in human GBM cells, various solid tumors, and various brain cancers and their corresponding normal adjacent tissues with hemoglobins (HB) being also expressed in human GBM cells (4,(22)(23)(24)(25). Recently, we have reported that MB variants 2,9,10,11,13 are differentially expressed in GBM cells (26). In continuation of our previous work, we sought to investigate MB expression and its correlation with hypoxia markers: CAIX and lactate dehydrogenase A (LDHA), in human tissue microarrays (TMA) of multiple organ tumors, different brain tumors, and GBM tissue samples and their relative cancer-adjacent normal tissues (CANT). Our data suggest that in different brain malignancies' including GBM, MB is regulated by hypoxia. We also detected an enhanced nuclear expression of LDHA in GBM. Contribution of these findings to the overall progression of different brain cancers, including GBM, is worthy of further exploration.

Statistical Analysis
Data are expressed as means ± standard error. Statistical analyses were performed using SigmaPlot 13 software (Systat Software Inc., Chicago, IL, USA). Differences between different tissues pathologies were assessed using One-way Analysis of Variance (ANOVA) then pairwise multiple comparison were made using Holm-Sidak method for multiple organ tumors TMA. For the other two TMA, different brain tumors and glioblastoma multiforme, multiple comparisons were made versus a control (normal brain tissue) using Dunn's test. Additionally, Spearman Rank correlation analysis was used to assess the relationship between MB expression and the tested markers of hypoxia, CAIX and LDHA. The level of significance was set at p <0.05.

RESULTS
We screened each group of TMA: multiple organ tumors (39 cases/43 cores), different brain tumors (40 cases/80 cores), and GBM tumors (40 cases/80 cores), each with its respective CANT for CAIX, LDHA, and MB protein expression. Subcellular distribution of these proteins and correlations of their expression levels in these samples were assessed.
Multiple organ tumor TMA contained samples from malignant tumors, benign tumors, hyperplasia and CANT ( Table 1). Positive MB immunostaining was observed in all benign tumors (meningioma, ovarian serous papillary cystadenoma and adenomas of the thyroid and breast) and hyperplastic samples from cirrhotic liver and prostate. MB staining was also positive, and in a majority of malignant tumors (72%), mainly in thyroid papillary carcinoma, esophagus squamous cell carcinoma, nasopharyngeal carcinoma, hepatocellular carcinoma, breast invasive ductal carcinoma, kidney clear cell carcinoma, ovary serous cystadenocarcinoma, testis seminoma, skeletal muscle pleomorphic rhabdomyosarcoma, malignant melanoma of thigh, squamous cell carcinoma of scalp as well as adenocarcinomas of prostate, stomach, colon, pancreatic duct and lung. Likewise, MB immunostaining was positive in most (94%) CANT samples, specifically, in brain, lung, breast, kidney, ovary, bone marrow, skeletal muscle, and skin as well as CANT samples from different gastrointestinal organs (Table 1). Where expressed, MB localized mainly in the cytoplasm with few samples showing weak nuclear staining, mainly in normal tissues adjacent to lung, breast, and endometrial cancers, hyperplastic prostate tissue, and malignant melanoma of thigh (Figures 1, 2). CAIX was positive only in a few malignant tumors (38%, mainly in nasopharyngeal carcinoma, lung adenocarcinoma and squamous cell carcinoma, kidney clear cell carcinoma, cervical squamous cell carcinoma, ovarian serous cystadenocarcinoma and malignant melanoma of thigh) and CANT (12.5%, mainly in stomach and skin) samples ( Table  1). It was mainly localized to the cell membrane with a complete absence of nuclear staining (Figures 1, 2).
Positive LDHA immunostaining was observed in all benign tumors and hyperplastic samples, and in the majority of malignant tumors (95%) and CANT (81%) samples ( Table 1). A great deal of overlap between LDHA and MB positive samples was seen in malignant tumors apart from lung Squamous cell carcinoma, adenocarcinoma endometrium, cervical squamous cell carcinoma, osteosarcoma of the femur where LDHA immunostaining was positive while MB was negative. Similarly, LDHA immunostaining was positive in almost all MB positive tissue samples except lung and ovary CANT samples where LDHA immunostaining was negative while MB was positive. Where positive, LDHA showed a pattern of localization that was somewhat similar to that of MB, being mainly observed in the cytoplasm with weak-moderate nuclear staining detected in most of the tested samples (Figures 1, 2).
Hypoxia markers (CAIX and LDHA) and MB protein expression was also investigated in a TMA of different brain tumors. The TMA contained tissue samples from malignant brain tumors (astrocytoma, protoplasmic astrocytoma, atypical astrocytoma, glioblastoma, oligodendroglioma, and anaplastic oligodendroglioma), CANT, and normal brain tissue. CAIX immunostaining was positive in the majority of malignant brain tumors (78%), but was reduced (10%) or absent from CANT and normal brain tissues, respectively ( Table 2). Where expressed, CAIX was mainly detected in the cytoplasm with few samples showing membranous localization (3 glioblastomas and 1 oligodendroglioma) and weak nuclear staining (2 astrocytomas) ( Figure 3). Significantly less cytoplasmic LDHA staining was observed in malignant tumors compared to normal brain tissue, which was equivalent to CANT (50% vs. 89% vs. 76%, respectively p = <0.001). This is in contrast to the significant increase in the percentage of cells with cytoplasmic CAIX staining observed in malignant brain tumors (25%) compared to normal brain tissue (0%) and CANT (3%) (p = <0.001). Although LDHA showed mainly cytoplasmic localization, weak-strong nuclear localization was also observed in malignant brain tumors, CANT, and normal brain tissues with no significant differences being detected in the percent of cells showing nuclear CAIX or LDHA staining ( Table 2 and Figure 3). MB protein expression was positive in the majority of malignant brain tumors (60%), and CANT (90%) and normal brain tissue samples (90%) ( Table 2). MB immunostaining was TABLE 1 | MB, CAIX, and LDHA expression in tissue microarrays of benign tumors, multiple organ tumors, hyperplasia, and cancer adjacent-normal tissue.

Organ
Pathology mainly cytoplasmic with weak-strong nuclear localization being also observed in malignant tumors, CANT, and normal brain tissue ( Figure 3). Next, we investigated CAIX, LDHA and MB protein expression in a TMA of GBM tumors, CANT, and normal brain tissue. CAIX immunostaining was positive in the majority of GBM samples but was absent from all CANT and normal brain tissue specimens ( Table 3). Where expressed, CAIX mainly showed cytoplasmic and membranous localization. Additionally, a majority of GBM (98%), CANT (100%) and normal brain (67%) tissue samples were positive for LDHA ( Table 3). Cytosolic localization of LDHA was observed in malignant, CANT and normal brain, however, the percentage of cells showing nuclear LDHA was significantly higher in GBM compared to normal brain (35% vs. 2%, p = <0.001, Figures 4, 5, Table 3) with CANT samples showing a higher mean value than normal brain (27% vs. 2%, respectively, p = 0.062, Figures 4, 5, Table 3). Finally, in the GBM TMA, there was no significant difference in the percentage cells positive for MB (cytoplasmic or nuclear) staining among the three tissue types. Immunostaining of MB, CAIX, and LDHA is shown in higher magnification in Figure 6).
In an effort to gain further insights into the regulation of tumor MB protein expression by hypoxia, Spearman Rank Order Correlation analysis of MB and both hypoxia markers (CAIX and LDHA) was carried out. Significant positive correlations between MB and LDHA (r = 0.355, p = 0.00549) and CAIX (r = 0.511, p = 0.0000354) immunostaining was observed in different malignancies of the brain (Figure 7). Similarly, MB immunostaining correlated positively with LDHA (r = 0.277, p = 0.00999) and CAIX (r = 0.286, p = 0.00772) in GBM samples ( Figure 7). These results suggest that MB expression might be regulated by hypoxia in different brain malignancies including GBM. Intriguingly, MB immunostaining intensity showed a significant positive correlation with tumor grade in different brain malignancies (r = 0.553, p = 0.0000207, Figure 8). Collectively, these results suggest that MB is regulated, at least in part, by hypoxia and that its expression in different brain malignancies, including GBM, may be a marker of a more aggressive cancer phenotype as indicated by its positive correlation with tumor grade in different brain cancers.

DISCUSSION
Globins are expressed in different tumors and their expression has been shown to be altered under low oxygen tensions (4, 17,  19, 22,24,26). For example, it has been shown that Ngb and Cygb are expressed in GBM and are up-regulated by hypoxic conditions (4,22). These observations potentially explain GBM's remarkable ability to survive under hypoxia and the frequent recurrence rates associated with this cancer. Here, we confirm findings of other investigators that MB protein is expressed in a wide variety of cancers and we also report MB expression in some benign tumors, cancer-adjacent normal tissues, hyperplastic tissue samples and normal brain tissue. Our data also indicate that low oxygen tensions modulate MB protein expression in different brain cancers, including GBM. Enhanced nuclear LDHA immunoreactivity was readily detected in GBM; an observation that has not been previously reported. Finally, our results suggest that MB expression might serve as a marker of a more aggressive cancer phenotype, particularly in brain, as indicated by the positive correlation with the grade of different brain cancers.
Initially, we used a TMA containing a wide variety of cancers of different organs, benign tumors, CANT, and hyperplastic samples to generally assess MB expression and its correlation with markers of hypoxia. Using this highly variable array of samples, we could not detect a difference in MB protein expression or a correlation with known markers of hypoxia, CAIX and LDHA.  2 | MB, CAIX, and LDHA expression in tissue microarrays of normal brain tissues, brain tumors, and cancer adjacent normal tissue.

Pathology/Brain
Grade MB(C/N) LDHA(C/N) CAIX(C/N) Consistent with earlier reports (27), we detected no CAIX expression in any of the normal brain tissues. Also, similar to an earlier report, we observed weak to absent CAIX immunostaining in astrocytoma and oligodendroglioma samples with GBM samples showing somewhat more intense (weak-strong) CAIX staining that localized mainly to cell membrane (27). In our study, CAIX expression was significantly associated with a malignant phenotype which is consistent with observations of other researchers who reported that CAIX expression increases with tumor grade (27)(28)(29). Together, these observations underline the contribution of this protein to the malignant phenotype.
LDHA localizes mainly in the cytoplasm exerting its lactateproducing function via the conversion of pyruvate to lactate and is up-regulated in many tumors (30,31). Total LDHA expression has shown potential as a prognostic marker in some cancers such as clear cell renal cell carcinoma and gastric cancer (31,32). Studies aimed at investigating subcellular localization of LDHA have reported cytoplasmic and nuclear localization of LDHA in some cancers such as esophageal squamous cell carcinoma. In fact in this cancer, high nuclear expression significantly correlated with advanced tumor stage and more lymph node metastases (33). The study also showed that the survival rate of patients with high expression of nuclear/cytoplasmic LDHA was significantly lower than that of patients with low expression with dual high expression of nuclear and cytoplasmic LDHA being associated with substantially lower survival rates than that of patients with single high or dual low nuclear and cytoplasmic LDHA. Interestingly, a recent article showed that reactive oxygen species (ROS) promote LDHA nuclear translocation in an immortalized human keratinocyte (HaCaT), human cervical cancer (HT-3), human bone osteosarcoma epithelial (U2OS), and HeLa cell lines where LDHA gains a noncanonical enzyme activity producing the antioxidant metabolite, ahydroxybutyrate, potentially to protect cancer cells from ROSmediated cell death (34). Consistent with these findings, in the different brain tumors TMA we observed a lower percent of cells with cytoplasmic LDHA staining compared to normal brain and a trend toward increased nuclear LDHA immunostaining in malignant brain tumors compared to normal brain. Further and in support of nuclear LDHA enhancement in malignant tumors, this pattern was also reproduced in our GBM TMA tissue microarray.
It should be noted that MB is expressed in normal human brain and other normal tissues (35)(36)(37) and different types of cancers and cell lines (17, 19-21, 26, 38-41). On the other hand, CAIX expression is restricted to a few normal tissues in the GIT (e.g. intrahepatic bile ducts, gastric mucosa, and duodenum) and is absent in the other normal tissues. It is differentially expressed in cancers with different expression patterns (e.g. diffuse, focal, membranous) (29,(42)(43)(44)(45)(46) and frequent expression is reported at the periphery of necrotic areas (42). Additionally, CAIX staining was not shown in all tissues of the same type of cancer or different type of cancers (43,46). For example, the percentage of CAIX (IHC) positive cases in 1229 different types of cancer cases was about 30%, with 0.0%, 12.0%, 29.0%, and 0.0% CAIX staining being reported for prostate adenoma and carcinoma, pancreatic carcinoma, lung adenocarcinoma, and thyroid follicular adenoma, respectively (46). In contrast to CAIX, LDHA is expressed in normal tissues (47) and in many types of cancers, hypoxia-induced its expression (48)(49)(50). Based on this, it is quite expected not to see identical sites in the IHC images for MB and CAIX especially in normal tissues. However, we do see overlapping/similar pattern of staining for MB and  CAIX in cancers and even more similar/identical sites were observed for MB and LDHA staining. Based on this, we assume that MB is partially regulated by hypoxia in brain cancers. MB's role in cancer is controversial with some reports advocate for a tumor suppressor role and others for a cancer survival strategy (19,39,51,52). Association between MB expression and prolonged survival was observed in head and neck squamous cell carcinomas and breast cancer while correlation with poor survival and higher tumor grading was observed in lung cancer (in patients with adenocarcinoma) and clear cell renal carcinoma patients, respectively (18,19,39,52). Using a forced expression approach in A549 human lung carcinoma cells, MB exerted a tumor suppressor role via  reduction of the hypoxic response in spheroid cultures in vitro and in xeno-transplant mice models (51). The authors showed that Mb-mediated tumor oxygenation was the primary factor that promoted differentiation of cancer cells and suppressed both local and distal metastatic spreading because these effects were not observed using mutated forms of MB unable to bind oxygen and they were abrogated by the constitutive expression of an active form of HIF-1a (51). Using MB loss-of-function approach in stable knockdown MDA-MB468 cell line, increased oxygen uptake and elevated activities of mitochondrial enzymes during hypoxia, in addition to decreased cell viability and motility under both normoxic and hypoxic conditions were observed suggesting that MB also possesses unconventional functions that are not directly linked to its oxygen binding and transporting capacity (40). The authors assumed that MB potentially act as a shuttle for fatty acids thus supporting lipogenesis and cellular growth even at times when the oxygen supply is non-limiting (40). Consistent with previous reports supporting MB's role in cancer cell survival, we not only detect significant positive correlations between MB and each of the tested markers of hypoxia in different brain cancers and in GBM, but we also detect a positive correlation between MB immunostaining and tumor grade ( Figures 5, 7). These results suggest that MB expression possibly mediates a more aggressive cancer phenotype, particularly in the brain. Of note, this report is the first to show MB protein expression in normal human brain with neuroglobin, cytoglobin and hemoglobin already known to be expressed in normal brain (53). In fact, it has long been known that non-muscle MB is expressed in brains of hypoxia tolerant species such as hypoxiatolerant carp, lungfish and subterranean mole rat Spalax (21,54,55) however its exact role is yet to be determined. Interestingly, it has been shown that the total Mb copy numbers in the P. annectens skeletal muscle and heart were lower than the brain, in the order of a tenth or less. Thus, in lungfish, it is considered that the brain is the main site of Mb expression not the heart or skeletal muscles and it was thought that MB might be vital in supplying oxygen to the nervous system during aestivation periods (54). Aside from human brain, we found reports showing MB protein expression in adult hematopoietic stem/progenitor cells obtained from peripheral or bone-marrow blood (56) and glandular cells of breast (18) and prostate (20). Studies aimed at exploring how human brain MB expression compares to human heart and skeletal muscles and its role in normal brain physiology should be useful additions to our current knowledge of MB function.
Similar to our results here, positive correlations between MB and different markers of hypoxia were observed in different cancers. For example, MB mRNA expression correlated with HIF1a (r = 0.483) or VEGFa (r = 0.468) in lung cancer tissue samples (52). In prostate cancer, MB immunostaining positively correlated with CAIX (r = 0.246) (20) and in invasive breast cancer MB immunostaining positively correlated with CAIX (r = 0.286) and HIF-2 a (r = 0.293) (18). Although these correlations are generally weak to moderate, they are in agreement with our reported correlation coefficients both in direction and in magnitude and indeed suggest a partial regulation of MB by hypoxia in a wide variety of cancers including different brain tumors and GBM. Nevertheless, they also indicate that additional factors drive MB expression in these cancers. Furthermore, it is conceivable that these factors are likely common among different cancers and are not dependent on the type of cancer in question.
A relatively recent study has demonstrated the existence of 19 splice variants for MB, of which variants 9, 10, 11, and 13 are protein-coding cancer-specific variants, and their expression was  confirmed in MDA-MB468 breast cancer and DLD-1 colon cancer cell lines as well as in breast cancer biopsies (38). In these cell lines, the standard muscle-type MB transcript (variant 2) was also expressed, however in contrast to the cancer-specific variants (variants 9, 10 and 11) which were significantly induced, its expression was not affected by hypoxia (38,40). Variants 9, 10, and 11 encode the standard MB protein as found in muscles (38). Based on this observation, it was assumed that the expression of these variants underlie the hypoxia-mediated increase in MB protein expression in breast cancer cell lines and tumor biopsies (38,40). This assumption was also backed up by results from tumor biopsies where no correlation between MB variant 2 and VEGFA was detected, while significant positive correlations between VEGFA mRNA and MB variants 9, 10, 11 mRNA were observed (38). Likewise, we assume that the positive correlations observed herein between MB and markers of hypoxia at the In summary we show that MB protein is expressed in a wide variety of cancers, benign tumors, cancer-adjacent normal tissues, hyperplastic tissue samples and normal brain tissue. Our data suggest that low oxygen tensions modulate MB protein expression in different brain cancers including GBM. We also detected enhanced nuclear immunoreactivity of LDHA in GBM that was not previously reported supporting a role for nuclear LDHA in promoting more aggressive tumor phenotype. Finally, MB expression is associated with a more aggressive cancer phenotype, specifically in brain

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
MEE: formal analysis, investigation, methodology, writing -original draft, and writing -review and editing. MS: methodology and writing -review and editing. RE-T: conceptualization, methodology, and writing -review and editing. IE: writing -review and editing. OH: data curation, methodology, visualization, and writing -review and editing. MM: data curation, methodology, visualization, and writing -review and editing. JA-T: conceptualization and writingreview and editing. ME: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, visualization, writingreview and editing. All authors contributed to the article and approved the submitted version.