Chromosome 8q24 amplification associated with human hepatocellular carcinoma predicts MYC/ZEB1/MIZ1 transcriptional regulation

Abstract

Genomic instability is associated with late stage carcinomas and the epithelial mesenchymal transition (EMT). Of note is chromosome 8q24 amplification that has been documented in many epithelial-derived carcinomas. On this amplified region is the potent oncogene, c-myc. Not only does MYC overexpression activate targets that promote cell proliferation, it also activates transcription factors that drive EMT, including ZEB1. Further reinforcing EMT, overexpressed MYC also represses tumor suppressors involved in promoting the epithelial phenotype, including MIZ1. We predict that as carcinomas progress, chromosome 8q24 is amplified leading to high MYC levels that leads to ZEB1 expression and MIZ1 repression driving cells through EMT. To interrogate this clinically, limited cohorts of human epithelial-derived carcinomas were examined for MYC/ZEB1/MIZ1 expression patterns across increasing carcinoma grades. Interestingly, the predicted temporal patterns were only observed in hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinomas. Yet MIZ1 proved to be an excellent marker to assess carcinoma progression across types. We expanded the HCC cohort and determined that c-myc amplification was restricted to grade III/IV HCC that also exhibited increased MYC and ZEB1 nuclear expression whereas cytosolic MIZ1 expression was lost and only nuclear expression retained. These same resections were obtained from only individuals who had histories of alcohol consumption that were also diagnosed with cirrhosis, metastasis and had viral hepatitis suggesting etiology-specific mechanisms of cancer progression. Finally, analysis performed in Hep3B cells determined that alterations in MYC expression promoted the predicted changes in ZEB1 and MIZ1 expression and/or distributions and in markers for EMT further suggesting a relationship among these three transcription factors in HCC and their correlation to driving EMT.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-75219-1.

Introduction

Hepatocellular carcinoma (HCC) is the fifth most common cancer world-wide with over 600,000 new cases/year and a dismal five-year survival rate of less than 9%. Most HCC patients have chronic liver disease resulting from HCV/HBV infection, chronic alcohol consumption, aflatoxin exposure and/or fatty liver. Upon hepatic injury and the development of cirrhosis, hepatocytes become hyperplastic, then dysplastic and an invasive phenotype is acquired. This phenotype is characterized by a highly disorganized and rigid extracellular matrix that is distinct from that associated with fibrosis¹. This modified matrix promotes neoplastic transformation and the epithelial-mesenchymal transition (EMT)¹. As cells undergo EMT, epithelial apical-basolateral polarity is lost to allow for directed cell migration and metastasis^2–5. Cancer cells rarely become fully mesenchymal and instead display hybrid phenotypes characterized by cell plasticity. EMT is also associated with the acquisition of stemness where the cancer stem cells either self-renew or undergo a mesenchymal to epithelial transition (MET) to a more epithelial phenotype^2–5. More recent studies have further suggested that re-acquisition of aspects of the polarized epithelial phenotype allows metastatic lesions to establish at secondary sites. Because EMT is linked to drug resistance, understanding the mechanisms responsible for loss/reacquisition of polarity is critical to fully understand malignant transformation and to diagnose and treat HCC.

Genomic instability is associated with the later stages of HCC progression and EMT. Of note is chromosome 8q24 amplification that has been documented in many epithelial-derived carcinomas including HCC^6,7. Limited studies in HCC derived from different forms of livery injury further suggest that chromosome 8q24 amplification may be etiology-dependent. For example, 8q24 amplification was found to be more prevalent in HCC associated with alcohol-induced liver injury or from HCV/HBV infection^8,9 Although a subset (5/11) of metabolic dysfunction-associated steatohepatitis (MASH-)-induced HCC cases were positive for c-myc gains¹⁰, 8q24 amplification is not thought to be associated with HCC that arises from metabolic dysfunction-associated steatotic liver disease (MASLD) or aflatoxin exposure^8,9 On this amplified region of 8q24 is the potent oncogene, c-myc (at 8q24.21). It is widely appreciated that MYC overexpression leads to the transcription of countless targets that promote cell proliferation^11,12 However, more recently, MYC has been described as a master regulator of EMT and metastasis by promoting the expression of the transcription factors that drive EMT, including zinc finger E-box binding homeobox 1 (ZEB1)^13,14. ZEB1’s function in EMT is largely linked to its repression of cell polarity and the epithelial phenotype^13,15,16. Moreover, decreased ZEB1 levels have been linked to MET and the reacquisition of polarity¹⁵. Out of the few transcription factors that drive EMT, ZEB1 is also the most highly associated with drug resistance¹⁵. Further reinforcing progression through EMT, MYC also represses other transcription factors that function as tumor suppressors, including MYC-interacting zinc-finger protein 1 (MIZ1)^11,17. In normal cells, MYC-MAX dimers activate their countless targets while MIZ1 at its core promoter activates genes involved in promoting the epithelial phenotype, including those that function in adhesion, autophagy, apoptosis and polarity. In cancer cells with high MYC levels, MYC-MAX complexes are additionally recruited to MIZ1 at its promoter thereby repressing MIZ1 transcriptional activity leading to decreased adhesion, apoptosis, autophagy, and polarity and thus, a loss of the epithelial phenotype^11,12,17.

The connection between MYC overexpression and the corresponding changes in ZEB1 and MIZ1 have been reported in other cancer types, but generally not all three together. Nor have these relationships been extensively characterized in the context of HCC across grades or with respect to etiologies—especially not MIZ1. Thus, we propose that alcohol- and/or hepatitis-induced high grade III/IV HCC cases are characterized by chromosome 8q24 amplification that leads to high MYC levels. The high MYC leads to MIZ1 transcriptional repression and loss of the epithelial phenotype. Concomitantly, the high MYC activates ZEB1 expression driving cells through EMT by increasing mesenchymal/stemness traits while repressing the epithelial phenotype. To interrogate this scenario, we established the temporal expression patterns of MYC, ZEB1 and MIZ1 in human HCC resections and compared them to resections from other epithelial-derived carcinomas using immunohistochemistry (IHC). The results from both HCC and intrahepatic cholangiocarcinoma (ICC) are consistent with the prediction. We expanded the cohort for HCC samples and further confirmed the temporal expression patterns and confirmed c-myc amplification by fluorescence in-situ hybridization (FISH). When the temporal expression patterns were compared to patient history data retrospectively, we found that of the high grade cancers that exhibited amplified c-myc, high MYC and ZEB1 expression and decreased cytosolic MIZ1 with only nuclear MIZ1 labeling, nearly all were associated with metastasis and only to patients that had a history of both alcohol consumption and viral hepatitis. Analysis performed in Hep3B cells determined that alterations in MYC expression promoted the predicted corresponding changes in ZEB1 and MIZ1 expression and/or distributions and in markers for EMT further confirming a relationship among these three transcription factors in HCC and their correlation to driving EMT.

Results

Evidence for c-myc amplification and subsequent ZEB1 and MIZ1 transcriptional regulation is apparent only in HCC and intra-hepatic cholangiocarcinomas

To initially test our proposed scenario that in high HCC grades III/IV, c-myc is amplified leading to high MYC levels that in turn promote MIZ1 repression and increased ZEB1 expression, we examined a limited cohort of increasing grades of HCC cases (total n = 26). Four µm thick sections of formalin-fixed, paraffin-embedded tissue containing adjacent benign and malignant tumor lesions were immunohistochemically stained for MYC, ZEB1 and MIZ1. When scored for expression, the predicted temporal staining patterns were observed with high MYC and ZEB1 in the majority of grade III and IV cases with a dramatic decrease in cytoplasmic MIZ1 levels in grade IV cases (Table 1). Interestingly, MIZ1 labeling was restricted to the nucleus in these grade IV cases (explained in detail below). Because intrahepatic cholangiocarcinoma (ICC) has also been associated with 8q24 amplification^18,19, a similar sized ICC cohort (total n = 23) was examined for the MYC/ZEB1/MIZ1 temporal expression patterns. As for HCC, the predicted staining patterns were observed with high MYC and ZEB1 detected in the majority of grade III ICC cases (60% positive cases) with a concomitant decrease in cytoplasmic MIZ1 labeling (30% positive cases) and retained nuclear-only labeling of MIZ1 (Table 1).

Table 1.

Only hepatocellular carcinoma and cholangiocarcinoma cases are consistent with c-myc amplification and subsequent transcriptional regulation of ZEB1 and MIZ1.

Carcinoma	n	MYC nuclear (% positive)	ZEB1 nuclear (% positive)	MIZ1 cytoplasmic (% positive)	MIZ1 nuclear only (% positive)
Hepatocellular carcinoma
Grade I	4	–	–	4 (100%)	–
Grade II	10	2 (20%)	–	10 (100%)	–
Grade III	8	4 (50%)	3 (38%)	3 (38%)	3 (38%)
Grade IV	4	3 (75%)	3 (75%)	–	3 (75%)
Cholangiocarcinoma
Grade I	3	–	–	3 (100%)	–
Grade II	10	2 (20%)	3 (30%)	10 (100%)	2 (20%)
Grade III	10	6 (60%)	6 (60%)	3 (30%)	3 (30%)
Renal cell carcinoma
Grade I	4	–	3 (75%)	4 (100%)	–
Grade II	9	2 (22%)	–	9 (100%)	2 (22%)
Grade III	7	2 (29%)	–	2 (29%)	2 (29%)
Breast carcinoma
Grade I	6	–	2 (33%)	6 (100%)	1 (17%)
Grade II	13	1 (8%)	3 (23%)	13 (100%)	2 (15%)
Grade III	11	3 (27%)	3 (27%)	4 (36%)	1 (9%)
Pancreatic carcinoma
Grade I	6	–	2 (33%)	6 (100%)	–
Grade II	14	2 (14%)	3 (21%)	14 (100%)	1 (7%)
Grade III	11	3 (27%)	3 (27%)	5 (45%)	2 (18%)
Colon carcinoma
Grade I	7	–	–	7 (100%)	–
Grade II	13	2 (15%)	1 (7%)	13 (100%)	3 (23%)
Grade III	12	2 (17%)	6 (30%)	5 (41%)	2 (17%)

As a negative control, we examined renal cell carcinoma cases (RCC) (total n = 20). Importantly, chromosome 8q24 amplification is not associated with this cancer type. Although MIZ1 cytoplasmic labeling was decreased in late-grade III RCC, no dramatic increase in MYC labeling was observed while ZEB1 expression was absent. Instead, ZEB1 labeling was surprisingly observed only in grade I cases. Additionally, no cases that were positive for MYC were positive for ZEB1. For comparison, we also examined limited cohorts of other epithelial-derived carcinomas from breast (n = 30), pancreas (n = 31) and colon (n = 32). Although chromosome 8q24 amplification has been reported for these carcinomas, it is not as highly associated as are other genomic instabilities^6,7. However, no substantial increases in MYC were observed in high-grade cases for any of these carcinomas (Table 1). ZEB1 expression levels also did not change in increasing grades of breast or pancreatic carcinomas and only a modest increase was observed in grade III colorectal carcinoma. Although not quite as striking as for HCC and ICC, the MIZ1 labeling decreased in increasing grades of these carcinomas with a subset of high grade cases with nuclear-only MIZ1 labeling. Together these data suggest little to no chromosome 8q24 amplification in these cases and that its amplification may be more prevalent in liver-derived cancers.

Tissue sections of the indicated carcinomas and grades were immunohistochemically stained for MYC, ZEB1 and MIZ1. The number of samples positive for each is given and the percent of total is in parentheses. For MIZ1, cytoplasmic or nuclear only expression was scored separately.

The c-myc gene is amplified in high grade malignant HCC tumors

To confirm chromosome 8q24 and c-myc (at 8q24.21) were amplified in the high grade HCC tumors, we performed FISH analysis on ten grade III and eight grade IV cases. Images of hematoxylin and eosin stained resections at low magnification identified the benign (B) and malignant (M) components of each sample (Fig. 1A, a). When viewed at an intermediate magnification, the benign component displayed polarized hepatic morphology (Fig. 1A, b) whereas the malignant component was characterized by disorganized and dysmorphic cells (Fig. 1A, c) which was even more apparent at higher magnification (Fig. 1A, d). In particular, these cells displayed moderate to marked nuclear polymorphism with enlarged nucleolii and exhibited a mid/macro trabecular tissue pattern. Sections were labeled for Ki-67 to confirm increased cellular proliferation. As predicted, Ki-67 levels were low in the benign component of each section (Fig. 1A, e and f), but detected at high levels in the malignancy (Fig. 1A, g). Higher magnification confirmed Ki-67 nuclear localization (Fig. 1A, h). We also confirmed high MYC expression in the selected cases. As for Ki-67, little to no MYC was detected in the benign tissue (Fig. 1A, I and j) whereas high MYC was detected in the malignant cells (Fig. 1A, k) that was restricted to the nucleus (Fig. 1A, l). In the malignant portions of the sections, MYC labeling was scored as diffuse (score = 3) and intense (score = 3) with a composite score of 6 confirming high MYC expression (and see Fig. 2).

Fig. 1.

The c-myc gene is amplified in high grade malignant HCC tumors. A, Serial sections from the same resection were hematoxylin and eosin stained (a-d), labeled for Ki-67 (e-h) or MYC (i-l). Low, intermediate and high magnification images are shown as indicated. In the low magnification images, the benign (B) and malignant (M) components are apparent. In the high magnification images of the malignant tissue, dysmorphic cells are apparent (d) and nuclear Ki-67 and MYC labeling are observed (h and l, respectively). B, Serial sections from the same HCC resection were processed for FISH labeling of chromosome 8 centromere (control, in green) and c-myc (in red). Nuclei were stained with DAPI. Merged images are shown that reveal the c-myc positive puncta outnumber the chromosome 8 centromere positive puncta in the malignant nuclei, indicating c-myc amplification. C, The number of chromosome 8 centromeres were counted/nuclei. Values shown represent % of total nuclei counted with nuclei containing 0–2 centromeres or > 2 (ranging from 3–8). Also shown are the ratios of c-myc puncta vs. centromere puncta indicating focal amplification in the c-myc overexpressing nuclei. D, Benign and malignant RCC resections that displayed high MYC levels with no ZEB1 expression (a and b) or with high ZEB1 and no MYC expression (c and d) were processed for FISH labeling of chromosome 8 centromere (control, in green) and c-myc (in red). Nuclei were stained with DAPI. Merged images are shown with equal numbers of c-myc positive and chromosome 8 centromere positive in nuclei of both benign and malignant cells indicating no c-myc amplification.

Fig. 2.

High MYC nuclear expression is observed in high grade III and IV HCC only. A, Increasing grades of HCC tumors were labeled for MYC. Low, intermediate and high magnification images are shown as indicated. In the low magnification images, the benign (B) and malignant (M) components are apparent. In the high magnification images, nuclear MYC labeling is detected in grade III and IV tumors only. The samples were scored for MYC (B) or Ki-67 (C) immunoreactivity for each tumor grade. Values are plotted as the percent positive of total for each grade. D, The labeling for MYC was further scored for intensity and percent positivity and composite scores assigned (see “Methods”). The number of samples with each composite score for each grade is provided with the percent of samples that had that score. Dashes indicate 0 samples/0% positive.

Serial sections from the same samples were labeled with probes specific for c-myc and the chromosome 8 centromere. The adjacent benign tissue in all cases tested showed virtually no chromosome duplication or c-myc amplification (Fig. 1B, a). In almost all nuclei examined, the same number of red (c-myc) and green (chromosome 8 centromere) nuclear puncta were observed with 0–2 chromosome 8 centromeres present in each nucleus. Only 2% of nuclei from benign tissues labeled for > 2 chromosome 8 centromeres (Fig. 1C). In contrast, the malignant tissue showed dramatic gene duplication and c-myc amplification with multiple c-myc-positive puncta observed in each nucleus (Fig. 1B, b and c). When viewed at higher magnification, increased red c-myc fluorescent nuclear puncta far out-number those in green in the malignant cell nuclei (Fig. 1B, c). When quantitated, the ratio of c-myc to chromosome 8 centromere labeling in the benign tissue was 1.2 consistent with no 8q24 amplification. In contrast, 5/10 (50%) high-grade III and 7/8 (88%) high-grade IV cases showed striking duplication with more than 70% of the nuclei positive for > 2 chromosome 8 centromeres (Fig. 1C). The ratios of c-myc to chromosome 8 centromere labeling ranged from 4.3 to 12.4 indicating that significant focal 8q24 amplification had also occurred in these samples. Importantly, these twelve cases with c-myc gene amplification, also exhibited amplification in > 90% malignant cells and high MYC expression by IHC. Together, these results indicate that high levels of MYC expression correlate with chromosome 8 duplication and the focal amplification of c-myc on chromosome 8q24.

As a negative control, we examined chromosome 8 duplication and focal c-myc amplification in resections from RCC that displayed high MYC levels with no ZEB1 expression or with high ZEB1 and no MYC expression using the same control and c-myc probes. As predicted, in the adjacent benign component from each resection, the same number of green chromosome 8 centromere puncta and red c-myc puncta were observed (Fig. 1D, a and c). Similar results were observed in the malignant high-grade RCC tissue section with high MYC expression and the low-grade RCC tissue with high ZEB1 expression indicating no c-myc amplification or chromosome 8 duplication (Fig. 1D, b and d).

Increased MYC and ZEB1 expression inversely correlates with decreased cytosolic MIZ1 expression in high grade malignant HCC tumors

Based on our proposed scenario, the prediction is that the high MYC levels in high grade HCC tumors should correspond to increased ZEB1 expression and decreased MIZ1 expression due to its repression by MYC at its core promoters. To test this prediction, we extended our initial cohort sample to 147 HCC resections and assessed MYC, ZEB1 and MIZ1 expression levels. Of the cases examined, 14 cases were low-grade I cases, 85 intermediate-grade II, 33 high-grade III and 15 high-grade IV cases.

From low magnifications of the entire resections, the benign (B) and malignant (M) components of each sample were identified (Fig. 2A, a–d). When viewed at intermediate magnification, little to no MYC expression was detected in low-grade I (Fig. 2A, e and i) or intermediate-grade II malignancies (Fig. 2A, f and j) whereas enhanced labeling was observed in high-grade III (Fig. 2A, g) and IV tumors (Fig. 2A, h). Nuclear MYC labeling was confirmed at higher magnification in the higher grade tumors (Fig. 2A, k and l). When quantitated, MYC labeling was detected in only 7% (1/14) low-grade I and 9% (8/85) intermediate-grade II cases (Fig. 2B) whereas expression was detected in 30% (10/33) high-grade III and 53% (8/15) high-grade IV cases. The same resections were also labeled for Ki-67 to assess cell proliferative state. Only 29% (4/14) of the grade I cases and 14% (12/85) of grade II cases were positive for Ki-67 labeling whereas 58% (19/33) grade III and 93% (14/15) grade IV cases were positive (Fig. 2C).

To confirm that increased levels of MYC were present in the higher grade cases, labeling was also scored for intensity and the extent of labeling. The intensity was scored as 0 (no staining), 1 (weak), 2 (moderate) or 3 (intense). Patterns were scored as 1 (focal labeling; 25% of cells positive), 2 (regional; ≥25% but < 50% of cells positive) or 3 (diffuse; ≥50% of cells positive). The extent and intensity tumor scores were added to give composite scores ranging from 0 to 6 for each specimen. The grade I case that was positive for MYC exhibited weak, focal nuclear expression (composite score 3) while the positive grade II cases were characterized by weak to moderate regional expression (composite scores of 3 to 4) (Fig. 2D). In contrast, high-grade III and IV tumors had enhanced staining intensities with diffuse nuclear expression detected (composite scores of 5 and 6) (Fig. 2D). For comparison, Ki-67 labeling in grade I cases displayed low focal expression (composite score 2) and in grade II labeling displayed low regional expression (composite score 3) whereas labeling was considered intense and diffuse (composite score 6) in high-grade III and IV cases.

We next examined ZEB1 expression and distributions in serial sections from the same resection shown in Fig. 1A. As predicted, little to no ZEB1 labeling was detected in the benign tissue (Fig. 3A, a and b) whereas high ZEB1 expression was observed in the malignant cells (Fig. 3A, c). As for MYC, higher magnification images revealed ZEB1 was restricted to the nucleus (Fig. 3A, d). ZEB1 exhibited a similar temporal expression pattern as MYC with expression observed in only higher grade III and IV tumors (Fig. 3B, f and g). Higher magnification further confirmed ZEB1 expression was restricted to the nucleus (Fig. 3B, j and k). When scored, no ZEB1 was detected in low-grade I tumors and only 1/85 grade II tumors (Fig. 3C) with a composite score of 3 (Fig. 3D). In contrast, 17.5% (6/33) high-grade III and 47% (7/15) high-grade IV cases (Fig. 3C) exhibited intense and diffuse staining patterns (composite scores of 5 and 6) (Fig. 3D). Importantly and consistent with our predicted scenario, all but one of the 13 high-grade cases that were positive for ZEB1 were also positive for MYC.

Fig. 3.

High ZEB1 expression is observed primarily in Grade III and IV HCC. A, Tissues were labeled for ZEB1. Low (a), intermediate (b and c) and high magnification images (d) are shown. In the low magnification images, the benign (B) and malignant (M) components are apparent. In the high magnification image of the malignant tissue, ZEB1 nuclear staining is observed. B, Increasing grades of HCC tumors were labeled for ZEB1. Low, intermediate and high magnification images are shown as indicated. In the low magnification images, the benign (B) and malignant (M) components are apparent. In the high magnification images, nuclear ZEB1 labeling is detected in grade III and IV tumors only. C, The samples were scored for ZEB1 immunoreactivity for each tumor grade. Values are plotted as the percent positive of total for each grade. D, The labeling for ZEB1 was further scored for intensity and percent positivity and composite scores assigned (see “Methods”). The number of samples with each composite score for each grade is provided with the percent of samples that had that score. Dashes indicate 0 samples/0% positive.

Serial sections from the same resection shown in Fig. 1A were also examined for MIZ1 expression. Consistent with our proposed scenario, high levels of MIZ1 were observed in the benign tissue (Fig. 4A, a and b). Unlike MYC and ZEB1, MIZ1 is a recycling transcription factor²⁰ and was detected at high levels in both the cytoplasm and nuclei (Fig. 4A, a and b). Also consistent with our proposed scenario, much less MIZ1 labeling was detected in the malignant tissue (Fig. 4A, c). When viewed at higher magnification, a near complete loss of MIZ1 cytoplasmic labeling was evident, however in some cases, nuclear labeling was retained (Fig. 4A, d). When monitored across tumor grades, robust cytosolic MIZ1 labeling was observed in grade I and II tumors (Fig. 4B, a, b, e, f). Higher magnification revealed that both nuclear and cytoplasmic labeling was detected (Fig. 4B, I, j). In contrast, the cytoplasmic labeling was dramatically decreased in grade III and IV tumors (Fig. 4, c, d, g, h). However, in a subset of the high grade tumors, MIZ1 expression was detected only in nuclei (an example is shown in Fig. 4B, h and l).

Fig. 4.

Decreased cytosolic MIZ1 expression inversely correlates with increased MYC and ZEB1 in high grade malignant HCC tumors. A, Tissues were labeled for MIZ1. Low (a), intermediate (b and c) and high magnification images (d) are shown. In the low magnification images, the benign (B) and malignant (M) components are apparent. High MIZ1 expression is observed in the benign tissue in both the cytoplasm and nucleus (a and b). The cytoplasmic labeling is dramatically reduced in the malignant component (c and d). In most cases, only nuclear MIZ1 labeling remains in the tumor. An example is shown in d. B, Increasing grades of HCC tumors were labeled for MIZ1. Low, intermediate and high magnification images are shown as indicated. In the low magnification images, the benign (B) and malignant (M) components are apparent. Cytoplasmic MIZ1 labeling is dramatically decreased in grade III and IV tumors. In most cases, only nuclear MIZ1 labeling is detected in grade III and IV tumors. An example of a tissue section devoid of both cytoplasmic and nuclear MIZ1 labeling is shown in g and k whereas an example with only nuclear MIZ1 labeling is shown in h and l. C, The samples were scored for MIZ1 cytoplasmic (C) or nuclear only (D) immunoreactivity for each tumor grade. Values are plotted as the percent positive of total for each HCC grade. MIZ1 cytoplasmic (E) or nuclear only labeling (F) was further scored for intensity and percent positivity and composite scores assigned (see “Methods”). The number of samples with each composite score for each grade is provided with the percent of samples that had that score. Dashes indicate 0 samples/0% positive.

When scored, we found that 100% low-grade I cases were positive for MIZ1 cytosolic labeling (Fig. 4C) and mainly showed moderate and diffuse expression with composite scores of 5 (Fig. 4D). As for grade I cases, 100% of grade II cases were positive for MIZ1 cytosolic labeling (Fig. 4C), but the composite scores were lower with the majority at 3 (Fig. 4E). MIZ1 cytoplasmic labeling was detected in only 36% (12/33) of grade III cases (Fig. 4C), but the composite scores were all low at 2 (Fig. 4E). However, of these 33 cases, five were found to exhibit high levels of nuclear labeling with scores of 4 and 5 (Fig. 4D and F). In contrast, all grade IV tumors were devoid of cytoplasmic labeling (15/15) whereas 60% (9/15) retained nuclear labeling (Fig. 4D and F) that was robust (composite scores of 5 and 6) (Fig. 4F). Remarkably, these same cases with nuclear-restricted MIZ1 expression also exhibited robust MYC and ZEB1 labeling and were positive for c-myc amplification. These findings are not only consistent with our proposed scenario, but they are also consistent with MIZ1 repression at its core promoter by increased MYC expression such that it was prevented from recycling and proteosomal degradation^21,22. Furthermore, the morphology of all these samples was characterized by marked nuclear polymorphism and enlarged nucleoli. The tissues also all displayed trabecular growth patterns with chords of ten or more cells thick. No glandular or scirrhous phenotypes were observed.

High MYC, ZEB1 and MIZ1 nuclear expression in only those high grade metastatic cases diagnosed with cirrhosis and with histories of viral hepatitis and alcohol consumption

To further examine what HCC etiologies correlated with chromosome 8q24 amplification, the patient histories of the high grade III and IV tumors were examined retrospectively. For this analysis, only the 18 cases that displayed high MYC expression were considered. Of these 18 cases, 13 were positive for high ZEB1 expression and 14 for MIZ1 staining that was restricted to the nucleus. Interestingly, nuclear expression of all three transcription factors was observed in only those patients with histories of viral hepatitis and alcohol consumption and were diagnosed with cirrhosis and metastatic HCC (Table 2). To confirm that ZEB1 expression also predicts metastasis^13,15,16, patient lymph nodes from all ZEB1 positive cases identified in the extended cohort were examined for the presence of tumor cells. Notably, metastasis was identified in all cases (Table S1).

Table 2.

Of the HCC cases with c-myc amplification, all were from individuals that had a history of viral hepatitis and alcohol consumption and were diagnosed with cirrhosis.

Clinical pathology	MYC (n = 18)	ZEB1 (n = 13)	MIZ1 nuclear only (n = 14)	Ki-67 positivity (n = 33)
Viral hepatitis	17 (94%)	13 (100%)	13 (93%)	26 (79%)
Alcohol consumption	18 (100%)	12 (92%)	14 (100%)	31 (94%)
Cirrhosis	17 (94%)	12 (92%)	13 (93%)	25 (76%)
Metastasis	18 (100%)	13 (100%)	14 (100%)	32 (97%)

Grade III and IV cases with unequivocal c-myc amplification were scored for MYC, ZEB1 and MIZ1 positivity. The number of samples positive for each is given and the percent of total is in parentheses. For MIZ1 cytoplasmic or nuclear only expression was scored separately. The clinicopathologic variables were obtained from patient medical records mainly from postoperative pathological examination and compared retrospectively.

To confirm that ZEB1 expression also predicts patient drug resistance^13,15,16, we compared overall patient survival from the time of diagnosis with respect to tumor grade (Fig. 5). For patients diagnosed with grade I HCC, overall survival ranged from four to almost eight years. In all cases, patients were treated with only sorafenib (Table S2). If diagnosed with grade II HCC, the prognosis was worse, with survival for almost 6 years down to three (Table S3). The majority of patients (61%) received a single therapeutic (generally sorafenib), but the remaining 39% also received hepatic intra-arterial chemotherapy suggestive of increased resistance (Table S3). If the patient was diagnosed with grade III HCC, survival was dramatically decreased to 10–15 months, with the majority surviving from 12 to 14 months (Table S4). In almost all cases, patients received two drug regimens, indicative of increased resistance (Table S5). Of those cases with c-myc amplification and increased ZEB1 expression, all but one patient received two drug regimens and survival was shifted down to 10–13 months with the majority surviving only 11–13 months (Fig. 5A). Although the sample size is small, the trend suggests that the already highly resistant grade III malignancies may be even more resistant in patients with c-myc amplification and increased ZEB1 expression. However, the trend is not as apparent for patients diagnosed with grade IV malignancies. All but two patients received two treatments and survival was further diminished to only 4–9 months (Table S6 and Fig. 5B). Although decreased resistance was modestly apparent, the tumors with c-myc amplification and high ZEB1 expression may be differently resistant.

Fig. 5.

Patient survival is decreased in those with c-myc amplification, high MYC, ZEB1 and MIZ1 nuclear only expression. Patient survival from the time of diagnosis with respect to tumor grade at the time of diagnosis was retrieved from patient records. The percent survival was calculated and plotted for individuals diagnosed with grade III (A) or grade IV (B) HCC.

Enhanced MYC expression in intermediate stage human Hep3B cells promotes the loss of cytosolic MIZ1 distributions and EMT

The IHC analysis established the temporal relationship between MYC, ZEB1 and MIZ1, but it does not directly test for it. At present, there are no good animal models that recapitulate the genomic instability observed in human HCC²³. The often-used DEN/alcohol HCC mouse model provokes genetic mutations not observed in humans and has not been reported to induce 8q24 amplification. Due to these limitations, we turned to intermediate grade human HCC-derived Hep3B cells for proof-of-concept experiments using a pharmacological approach to manipulate MYC expression levels. We treated cells with selumetinib (a MAP kinase inhibitor that leads to decreased MYC expression) or EGF (to increase MYC expression). As anticipated in untreated cells, all three transcription factors were localized in the nucleus (Fig. 6A, a, d and g) while an additional cytoplasmic pool was identified for MIZ1 (Fig. 6A, g).

Fig. 6.

MIZ1 cytosolic distributions inversely correlate with MYC expression. A, Hep3B cells were treated with 1 µg/ml selumetinib (sel) for 4 h to inhibit MYC expression or with 100 ng/ml EGF for 4 h to enhance MYC expression and labeled with the indicated antibodies. From micrographs, the Hep3B cells positive for nuclear MYC (B) or ZEB1 (C) were counted and the percent of total positive for either marker was calculated and plotted. The number of Hep3B cells positive for MIZ1 nuclear or cytosolic (D) staining were counted and the percent of total was calculated and plotted. Values are expressed as the average SEM from at least three independent experiments; *** p 0.001. E, Hep3B cells were treated in the absence or presence of 1 µg/ml selumetinib (sel) or 100 ng/ml EGF for 4 h. Total cell lysates were immunoblotted with the indicated antibodies. Tubulin served as the loading control. Relative levels of MYC (F), ZEB1 (G) or MIZ1 (H) were determined by densitometric analysis of the immunoreactive bands normalized to total α-tubulin. Values are expressed as the average SEM from at least three independent experiments; *** p 0.001. I, Cytosolic (C) and nuclear (N) fractions were prepared from cells treated in the absence or presence of 1 µg/ml selumetinib (sel) or 100 ng/ml EGF for 4 h and immunoblotted for MIZ1. J, Relative levels of MIZ1 were determined by densitometric analysis of immunoreactive bands and the percent cytosolic MIZ1 for each condition was calculated. Values are expressed as the average SEM from at least three independent experiments; *** p 0.001. The original uncropped blots are presented in Fig. S1.

Cells treated with selumetinib displayed a drastic reduction in MYC labeling with ZEB1 labeling also somewhat decreased (Fig. 6A, b and e). In contrast, enhanced MIZ1 cytoplasmic labeling was observed and its nuclear labeling was more diffuse consistent with its activation (Fig. 6A, h)²⁴. These phenotypes are remarkably consistent with what was observed in the benign regions of the human resections by IHC. In cells treated with EGF, nuclear MYC and ZEB1 labeling was more robust while MIZ1 cytoplasmic labeling was lost and it was detected only in discrete nuclear foci consistent with its repression by MYC-MAX dimers at its core promoter (Fig. 6A, c, f and i)²⁴. This redistribution is also remarkably consistent with what was observed in the malignant human HCC tissue samples. When quantitated, the percent of cells exhibiting nuclear MYC labeling was dramatically and significantly decreased in cells treated with selumetinib (Fig. 6B) whereas modest decreases were observed for ZEB1 labeling (Fig. 6C). Although MIZ1 nuclear staining was present in nearly all of the treated cells, its cytoplasmic levels were dramatically and significantly decreased in cells treated with EGF (Fig. 6D) as was observed in human malignant lesions.

To confirm our morphological results, we immunoblotted whole cell lysates prepared from cells treated in the absence or presence of selumetinib or EGF (Fig. 6E). When quantitated and normalized to total α-tubulin, MYC levels were decreased 3.3-fold in selumetinib-treated cells and were increased 1.5-fold in EGF-treated cells (Fig. 6F). Modest changes in ZEB1 levels were observed with a slight decrease observed in selumetinib-treated cells and slight increase observed in EGF-treated cells (Fig. 6G), but neither change was found to be significant. No changes in MIZ1 expression levels were observed in treated cells (Fig. 6H). To assess MIZ1 redistribution, we prepared and immunoblotted cytosolic and nuclear fractions. MIZ1 cytosolic distributions were dramatically decreased in EGF-treated cells (Fig. 6I). When quantitated, MIZ1 cytosolic distributions were ~ 38% in control cells, were increased to over 40% in selumetinib-treated cells, but dropped to only 10% in EGF-treated cells consistent with its inactivation by increased MYC expression levels (Fig. 6J).

To determine whether alterations in MYC expression promoted EMT, we also labeled for vimentin and E-cadherin in control and treated cells. As expected, vimentin was present in these intermediate grade HCC cells (Fig. 7A, d) and E-cadherin was detected at modest levels (Fig. 7A, g). In cells treated with selumetinib with decreased MYC (Fig. 7A, b), a modest decrease in vimentin labeling was observed especially in the cell periphery (Fig. 7A, e) whereas enhanced E-cadherin labeling was observed at sites of cell-cell contact (Fig. 7A, h). Such modest changes were expected given the short, 4 h treatment time. The opposite was observed in EGF-treated cells with decreased MYC expression (Fig. 7A, c). Vimentin labeling was modestly enhanced especially in the perinuclear regions of the filament network (Fig. 7, f) whereas E-cadherin labeling was noticeably decreased at sites of cell-cell contact (Fig. 7A, i). Immunoblotting confirmed our morphological results (Fig. 7B). When quantitated and normalized to total α-tubulin, vimentin levels moderately decreased in selumetinib-treated cells and moderately increased in EGF-treated cells (Fig. 7C) whereas E-cadherin showed the opposite (Fig. 7D) with a significant increase in cells with diminished MYC expression. Together these results suggest that alterations in MYC expression promote the predicted corresponding changes in ZEB1 and MIZ1 expression and/or distributions and in markers for EMT further confirming a relationship among these three transcription factors in HCC and their correlation to driving EMT.

Fig. 7.

MYC expression correlates with EMT progression. A, Hep3B cells were treated with 1 µg/ml selumetinib (sel) for 4 h to inhibit MYC expression or with 100 ng/ml EGF for 4 h to enhance MYC expression and labeled with the indicated antibodies. B, Hep3B cells were treated in the absence or presence of 1 µg/ml selumetinib (sel) or 100 ng/ml EGF for 4 h. Total cell lysates were immunoblotted with the indicated antibodies. Tubulin served as the loading control. Relative levels of vimentin (C) or E-cadherin (D) were determined by densitometric analysis of the immunoreactive bands normalized to total α-tubulin. Values are expressed as the average SEM from at least three independent experiments; *** p 0.001. The original uncropped blots are presented in Fig. S2.

Discussion

We determined that chromosome 8 duplication and focal c-myc amplification on chromosome 8q24 occurs exclusively in grade III and IV HCC which correlates with increased nuclear MYC expression. In these samples, nuclear ZEB1 expression was also enhanced whereas MIZ1 cytosolic expression was no longer detected and only nuclear expression retained. Smaller cohort sample analysis further revealed that this same MYC/ZEB1/MIZ1 temporal expression pattern was observed for high grade ICC, another carcinoma type associated with chromosome 8q24 amplification. However, and interestingly, this expression pattern was not observed in increasing grades of breast, colon or pancreatic carcinomas, cancer types that have been more loosely linked to 8q24 amplification. A larger HCC cohort analysis further revealed that in the grade III and IV samples with c-myc amplification, MYC and ZEB1 expression was also high with decreased MIZ1 expression. Additionally, nearly all of the same samples were from individuals who had a history of viral hepatitis and alcohol consumption and were diagnosed with cirrhosis and metastasis. Analysis performed in Hep3B cells determined that alterations in MYC expression levels promoted the predicted corresponding changes in ZEB1 and MIZ1 expression and/or distributions and in vimentin and E-cadherin expression further suggesting a direct relationship among these three transcription factors in HCC EMT.

Chromosome 8q24 amplification in liver-derived carcinomas

Chromosome 8q24 amplification has been documented in a number of epithelial cell-derived cancers besides HCC, including ICC and carcinomas from breast, colon and pancreas^6,7. However, we found evidence for prevalent c-myc amplification only in ICC and HCC corresponding with increased MYC levels. We also confirmed c-myc focal amplification in addition to chromosome 8 duplication in HCC and ICC (manuscript in preparation). In the other carcinomas examined, no significant increases in MYC expression were observed in high grade carcinomas and ZEB1 levels did not vary across grades. Since the MYC and ZEB1-positive RCC samples (our negative control) displayed no 8q24 amplification or chromosome 8 duplication, the increased MYC levels likely result from other regulatory mechanisms. However, MIZ1 cytosolic levels were consistently decreased in increasing grades of all carcinomas examined, but not to the extent as observed for HCC. Such consistent down regulation suggests that MIZ1 may be a highly universal regulator of polarity and serve as a good marker to assess the progression of various carcinomas. Interestingly, MIZ1 nuclear only labeling was observed in limited numbers in high grade breast, colon and pancreatic carcinomas. It will be interesting to test if these samples are also positive for increased c-myc amplification. Nonetheless, our results suggest that specific injuries associated with liver-derived cancers may promote selective genomic instability that promotes 8q24 amplification.

Most HCC patients have chronic liver disease resulting from HCV/HBV infection, chronic alcohol consumption, aflatoxin exposure, MASLD or MASH. Although most patients that develop HCC have cirrhosis, the mechanisms promoting HCC differ between pathologies^25–28. For example, HBV integrates into the host genome promoting specific mutations/alterations. Both HBV and HCV usurp host signaling pathways while aflatoxin metabolites promote specific DNA adducts leading to p53 mutations. Ethanol metabolism promotes the production of acetaldehyde, reactive lipid peroxidation products and reactive oxygen species that also promote DNA adducts and altered gene expression. The inflammation associated with alcohol-induced injury and hepatitis also promotes continuous cycles of necrosis and regeneration further promoting specific genetic alterations. Such differential changes in gene expression lead to varied alterations in DNA damage repair mechanisms, in cell cycle progression regulation (notably p53 loss of function associated with HCC), telomere shortening and chromosomal segregation that have all been associated with etiology-specific changes in genomic stability²⁵. Previous studies using array-based comparative genomic hybridization suggested that HCC due to ethanol exposure is more highly associated with 8q24 amplification than from HCV/HBV infection but not observed in HCC derived from MAFLD or aflatoxin^8,9. Studies using multiplex FISH analyses also confirmed that MASLD-induced-HCC did not display enhanced c-myc levels, but a subset of MASH-induced HCC cases (5/11) did¹⁰. However, our results predict that liver injury resulting from both viral hepatitis and ethanol exposure promote 8q24 amplification and metastasis. Because chromosome amplifications are not precise and of varying lengths, the cases that we identified by monitoring only c-myc amplification may represent a subset of the cases reported in the previous studies.

Chromosome 8q24 amplification predicts MYC/ZEB1/MIZ1 transcriptional regulation

Current models for cancer progression have suggested that EMT/MET transitions are highly dynamic processes. In general, normal polarized epithelial cells are nonproliferative, respond to apoptotic signals and are highly drug sensitive. As the epithelial cells acquire an invasive phenotype, they undergo an EMT where apical-basolateral polarity is lost and redirected to front-rear polarity to allow for directed cell migration and metastasis^2–5 (Fig. 8A). The mesenchymal cells are also characterized by drug resistance, invasiveness and immune evasion. However, cancer cells rarely become fully mesenchymal, but instead display hybrid phenotypes characterized by cell plasticity, the capacity to initiate tumors and metabolic adaptation. EMT is also associated with the acquisition of stemness where the cancer stem cells either self-renew or undergo MET to a more epithelial phenotype^2–5. More recent studies have further suggested that re-acquisition of aspects of the polarized epithelial phenotype (MET) is critical for establishing metastatic lesions at secondary sites (Fig. 8A).

Fig. 8.

EMT and MET in HCC cells are dynamic processes. A, More epithelial-like carcinoma cells have highest drug sensitivity, are nonproliferative and respond to apoptotic signaling whereas more mesenchymal-like cells are more proliferative, have highest drug resistance, invasion and immune evasion. A hybrid EMT state provides a plasticity window where cells can display stemness, can initiate tumor formation and can adapt to changing environments. B, Our proposed scenario for transcriptional dysregulation in HCC. In the normal polarized hepatocyte, high levels of MIZ1 are expressed (promoting polarity) whereas low levels of MYC and ZEB1 are expressed (reflecting decreased proliferation). Upon hepatic injury and the development of cirrhosis, hepatocytes become hyperplastic, then dysplastic, and as the invasive phenotype is acquired, increased chromosome instability at grades III and IV HCC is observed exemplified by c-myc amplification and increased MYC. The high Myc leads to MIZ1 transcriptional repression and loss of the epithelial phenotype. Concomitantly, the high MYC activates ZEB1 expression driving cells through EMT by increasing mesenchymal/stemness while repressing the epithelial phenotype. We further propose that as cells metastasize and colonize at secondary sites, epithelial traits are partially reacquired through the inactivation of ZEB1/MYC and reactivation of MIZ1.

Upon hepatic injury and the development of cirrhosis, hepatocytes become hyperplastic, then dysplastic, and as the invasive phenotype is acquired, increased chromosome instability is observed (Fig. 8B). Our results suggest that grades III and IV HCC are characterized by such chromosome instability as exemplified by c-myc amplification and increased MYC in these samples. The high MYC leads to MIZ1 transcriptional repression and loss of the epithelial phenotype and acquisition of mesenchymal and stem cell traits. Concomitantly, the high MYC activates ZEB1 expression driving cells through EMT by increasing mesenchymal/stemness while repressing the epithelial phenotype. Our finding that all ZEB1-positive samples were from patients with metastases is consistent with this scenario. We further propose that as cells metastasize and colonize at secondary sites, epithelial traits are partially reacquired through the inactivation of ZEB1 and reactivation of MIZ1 (Fig. 8B). A hypothesis we are currently exploring.

Translational impact: actionable targets and treatment strategies

The majority of therapeutics used to treat HCC are inhibitors of tyrosine kinases, cyclin-dependent kinases and programmed cell death protein-1/ligand-1 (PD-1/PDL-1)³⁰. However, these options have proven unsatisfactory due to the rapid development of drug resistance and toxicity²⁹. Thus, there is tremendous need to explore other targets to develop effective combination therapies²⁹. Recent studies indicate that tipping the MYC-MIZ1 balance in gene expression is a feasible actionable target²². In normal cells, MIZ1 is rapidly ubiquitinated by HUWE1 (an E3 ubiquitin ligase) and targeted for proteosomal degradation. Small molecules have been identified that inhibit HUWE1 (e.g., heclin, B18622, B18626)²² that promote MIZ1 stability, enhanced epithelialization and decreased cancer growth²⁰. Also exciting is the finding that these compounds synergize with clinically approved proteasome inhibitors (e.g., carfilzomib, bortezomib) to repress cancer progression. Also of note, are silibinin and honokiol, compounds that repress ZEB1 expression levels thereby inhibiting EMT and cancer cell stemness in bladder and breast cancer, respectively³⁰. Even more encouraging is that the FDA-approved class 1 HDAC inhibitor, mocetinostat, reverses ZEB1-mediated EMT and drug resistance in pancreatic cancer³¹. Such therapeutics should be considered in the context of treating HCC.

Methods

Research involving human samples

Studies involving human tissue samples were performed for research and investigational purposes only. The study was approved by the Institutional Review Board (IRB) of the Office of Research Oversight/Regulatory Affairs, Georgetown University. All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was obtained from all subjects and/or their legal guardian(s). Samples were obtained from the MedStar Georgetown University Hospital biobank.

Initial cohort selection

Small cohorts of carcinomas from liver, kidney, breast, pancreas and colon were selected. In all cases, formalin-fixed paraffin-embedded tissue sections containing adjacent benign and malignant tumors were examined. Patients selected for each cohort had histologic diagnosis of the respective carcinoma, underwent curative surgery with no presurgical treatment resulting in tissue necrosis. Patients with HCC (n = 26) or intra-hepatic cholangiocarcinoma (ICC) (n = 23) underwent curative hepatectomies and displayed no concurrent presence of another liver carcinoma. Breast carcinoma cases (n = 30) were from patients who underwent curative lobectomies. Pancreatic carcinoma cases (n = 31) were from patients who underwent curative pancreatectomies. Renal cell carcinoma (RCC) cases (n = 20) from patients who underwent curative nephrectomies and colorectal carcinoma cases (n = 32) were selected. For the breast, pancreatic, colon and RCC cases, all patients displayed no concurrent presence of another primary carcinoma with metastatic status.

Extended HCC cohort

One hundred forty-seven patients were included with the same inclusion criteria as described above. The clinicopathologic variables of viral hepatitis history, cirrhosis, tumor size, grade and number, vascular invasion and tumor-node-metastasis (TNM) stage were obtained from patient medical records mainly from postoperative pathological examination. The TNM staging was performed according to the seventh edition system³². The diagnosis of HCC was determined by examination of histologic morphology and expression of at least one marker of hepatocyte differentiation such as HepPar-1 and/or Arg-1. This HCC cohort contained 14 low-grade 1 (10%), 85 intermediate-grade 2 (58%), 48 high-grade cases with 33 high-grade III (22%), and 15 high-grade IV (10%) lesions. History of viral hepatitis was identified in 94 patients (64%), liver cirrhosis in 87 (59%), a history of alcohol consumption confirmed in 113 (77%) and metastasis was identified in 51 (35%) cases. The clinicopathologic features of the patient cohort are summarized in Table S6. Based on the total number of patients diagnosed with HCC at our institution during the above stated period vs. the number of patients that have undergone curative hepatectomies vs. the number of specimens which included excisional biopsies with both benign and malignant components present adjacently in the same block for direct side-by-side expression comparison, this extended HCC cohort has a power of analysis of P = 0.88 (88%).

Immunohistochemistry

Four-micron thick sections of the formalin-fixed, paraffin-embedded tissue blocks and tissue microarray slides were probed with antibodies against Ki-67 (clone Mib-1, Dako, Glostrup, Denmark), against MYC (mouse monoclonal clone Y69, Abcam, Boston, MA, dilution 1:50), against MIZ1 (rabbit polyclonal, Novus Biologicals, Centennial, CO, dilution 1:200), and against ZEB1 (mouse monoclonal clone OTI3G6, Abcam, dilution 1:100). For hepatic differentiation, cases were also evaluated for reactivity against Arg-1 (mouse monoclonal clone SL6ARG, Invitrogen, San Diego, CA, dilution 1:4000) and/or HepPar-1 (mouse monoclonal clone OCH1E5, Dako), using an automated method (DAKO EnVision + Dual Link System-HRP). Pretreatment of the formalin-fixed, paraffin-embedded tissue sections with heat-induced epitope retrieval (HIER) were performed using diluted Envision FLEX Target Retrieval Solution, Low pH (×50) (K8004). Deparaffinization, rehydration and epitope retrieval were performed in DAKO PT Link (PT100/PT101). The following parameters were used: pre-heat at 85℃, epitope retrieval at 97 °C for 20 min and cool down to 65 °C. Racks were placed in Envision Flex Wash Buffer (code K8007) for 5 min. The slides were treated with Flex Peroxidase Blocking solution (SM801) for 5 min, followed by incubation with primary antibodies for 20 min. The slides were treated sequentially with Flex Mouse Linker (SM804) for 15 min, Flex HRP (SM802) for 20 min and Flex diaminobenzidine (DAB) with Substrate-Chromogen (SM803) for 10 min. Sections were counterstained with hematoxylin for 5 min before being microscopically monitored. Normal liver tissues were used as positive control, while negative controls were performed by omitting the primary antibody. Non-liver high grade tumors were used as negative controls for HepPar-1, Arg-1 and MIZ1 staining. Cytoplasmic staining with or without nuclear staining in the tumor cells was considered positive. Positive nuclear staining for MYC, Ki-67, ZEB1 and positive cytoplasmic/nuclear staining for MIZ1 was classified as diffuse 3 (with ≥ 50% positive tumor cells were positive), regional 2 (with ≥ 25% but < 50% positive tumor cells) or focal 1 (with < 25% positive tumor cells). Negative immunoreactivity was characterized by the complete absence of MIZ1, MYC, ZEB1 expression or < 1% immunoreactivity for Ki-67. Label intensity was scored as 0 (no staining), 1 (weak), 2 (moderate), or 3 (intense). The extent and intensity scores were added to give a composite score (range: 0–6). Composite scores of 0–3 were considered low expression while scores of 4–6 were considered high expression. Immunoreactivity was semi-quantitatively scored in a blinded fashion by two pathologists.

Fluorescence in-situ hybridization

Commercial locus-specific probes for FITC-conjugated chromosome 8 centromere and Texas-Red-conjugated c-myc were used to monitor c-myc amplification (Dao-Agilent Technologies (Santa Clara, CA) using fluorescence in-situ hybridization (FISH). Before hybridization, formalin fixed paraffin embedded tissue sections were deparaffinized, pretreated for 10 min at 95–99 ± 2 °C in tissue pretreatment buffer and digested with cold pepsin at RT for 10 min. Slides were rinsed with wash buffer twice for 3 min, dehydrated with 70%, 85%, and 96% graded ethanol, and air dried before hybridization. The DNA probe mixture was applied to slides that were then sealed with a cover slip and rubber cement. The slides were placed in a pre-warmed humidified hybridization chamber followed by denaturation at 82 ± 2 °C for 10 min and hybridization at 45 ± 2 °C for 90 min. After coverslips were removed, slides were immersed in post-hybridization wash buffer at 63 ± 2 °C for 10 min. Slides were rinsed with wash buffer twice for 3 min, dehydrated with 70%, 85%, and 96% graded ethanol, and air dried before hybridization. Fluorescence mounting medium containing DAPI was added to the slide and the slide was sealed. Analysis was performed using the fully automated Bioview-Duet (Abbott, Abbott Park, IL) composed of an Olympus BX63 epifluorescence microscope (with 20, 40 and 60× objectives), equipped with a CCD camera (Olympus, Tokyo, Japan). Signals were evaluated at 400× and 600× magnification. Fifty morphologically unequivocal, non-overlapping neoplastic cells in the malignant component of each resected case were assessed for the presence of the c-myc and chromosome 8 centromere fluorescence. Scores were expressed as the actual copy numbers per cell in the majority of the cells in each sample. Amplification was scored when at least double the number of copies of c-myc were detected relative to the chromosome 8 centromere copy number in at least 20% of the analyzed cells. The scoring was evaluated with the observer blinded to the pathological details of the cases.

Cell culture

Hep3B cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Gemini Bio-Products, Woodland, CA). Cells were grown in a humidified 5% CO₂ incubator at 37°C. Hep3B cells were seeded onto glass coverslips in dishes at 0.5–1.0 × 10⁶ cells/dish and cultured for 1–2 days. Hep3B cells were treated with 1 µg/ml selumetinib (Selleckchem, Houston, TX) and 100 ng/ml EGF (Shenandoah Biotechnology, Warminster, PA) for 4 h in complete medium at 37 °C.

Immunofluorescence microscopy

Cells were fixed on ice with chilled phosphate-buffered saline (PBS) containing 4% paraformaldehyde for 1 min and permeabilized with ice-cold methanol for 10 min. Cells were processed for indirect immunofluorescence as previously described³³. Cells were labeled with rabbit polyclonal antibodies against MYC (Cell Signaling Technology, Danvers, MA), MIZ1, ZEB1, E-cadherin or vimentin (all four from Proteintech, Rosemont, IL). Alexa 488- or 568-conjugated secondary antibodies (Invitrogen Life Technologies, Carlsbad, CA) were used at 3–5 µg/ml. Labeled cells were visualized at RT by epifluorescence with an Olympus BX60 Fluorescence Microscope (OPELCO, Dulles, VA) using an UPlanFl 60×/NA 1.3, phase 3, oil immersion objective. Images were taken with an HQ2 CoolSnap digital camera (Roper Scientific, Germany) using MetaMorph software (Molecular Devices, Sunnyvale, CA). Adobe Photoshop (Adobe Systems, Mountain View, CA) was used to process images and to compile figures.

Preparation of nuclear and cytosolic fractions

Hep3B cells grown on 6 coverslips were scraped into 1 ml of 150 mM Hepes, pH 7.4, containing 1% NP-40 and protease inhibitors (2 µg/ml each of leupeptin, antipain, PMSF, and benzamidine). The homogenate was centrifuged for 5 min at 2,800 rpm at 4 °C and the supernatant was collected as the cytosolic fraction. The pellet was resuspended in 1 ml of the lysis buffer and centrifuged at 14,000 rpm for 3 min at 4 °C. The resultant nuclear fraction was resuspended to volume.

Immunoblotting

Cells were grown on coverslips, washed with chilled PBS, extracted with Laemmli sample buffer and boiled for 3 min³⁴. Proteins were electrophoretically separated using SDS-PAGE, transferred to nitrocellulose and immunoblotted with the rabbit polyclonal antibodies specific for MYC, MIZ1 or ZEB1 or mouse monoclonal antibodies specific for α-tubulin (Sigma Aldrich, St. Louis, MO). HRP-conjugated secondary antibodies were used at 5 ng/ml, and immunoreactivity was detected with enhanced chemiluminescence (Perkin Elmer, Crofton, MD) using the ChemiDoc Touch Imager (BioRad). Relative protein levels were determined by densitometric analysis of immunoreactive bands using ImageJ software (National Institutes of Health, Bethesda, MD).

Statistical analysis

Values are expressed as the mean ± SEM from at least three independent experiments. The One-way ANOVA test was used to analyze more than 2 groups of data, followed by Holm- Sidak multiple pairwise comparison. P values of ≤ 0.05 were considered significant.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

EMT

Epithelial mesenchymal transition

FISH

Fluorescence in-situ hybridization

HBV

Hepatitis B virus

HCC

Hepatocellular carcinoma

HCV

Hepatitis C virus

ICC

Iintrahepatic cholangiocarcinoma

IHC

Immunohistochemistry

MET

Mesenchymal to epithelial transition

MIZ1

Myc-interacting zinc-finger protein 1

MASLD

Metabolic dysfunction-associated steatotic liver disease

MASH

Metabolic dysfunction-associated steatohepatitis

RCC

Renal cell carcinoma cases

TNM

Tumor-node-metastasis

ZEB1

Zinc finger E-box binding homeobox 1

Author contributions

JJC, SSD and PLT designed the studies. JJC, SC, BVK and SSD performed experiments. All authors analyzed the data. JJC, SSD and PLT assembled figures and wrote the manuscript. All authors edited the manuscript.

Funding

These studies were supported by R01AA017626 and R21AA030353 awarded to PLT from the NIH/NIAAA.

Data availability

All data generated and analyzed during this study are included in this manuscript and supplementary files.

Competing interests

The authors declare no competing interests.