Aryl hydrocarbon receptor (AhR) is a ligand-activated transcription factor, which mediates the toxicity of dioxin and related compounds, and has an important role in development. However, a structural basis for ligand binding to the AhR remains unclear and the study was hindered by the low abundance and inherited instability of the AhR. Based on a previously defined minimal ligand-binding domain (LBD, residues 230-421), in the present study a series of truncated LBD constructs were created and expressed in insect cells (Sf9) using a baculovirus expression system. An antibody was produced to analyze the expressed. The antisera can detect as low as 0.3ng of AhR LBD from cytosol of Sf9. An in vitro [3H]TCDD binding assay was developed to characterized the expressed LBD. The assay yielded an estimate for the KD of C57Bl/6 mouse liver binding at 1.4nM. The present expression system yields soluble AhR LBD protein at ~0.15% of cytosol protein. Supplementation of the Sf9 culture medium with additional glucose resulted in an increase in the amount of soluble AhR, due to an increase in intracellular ATP level. However, cotransfection of LBD with hsp90 interaction protein p23 made no substantial change in the amount of cytosolic AhR. The soluble recombinant LBD retains functionality in the form of specific binding to dioxin, and its thermal stability was indistinguishable from that of mouse liver. However the ligand-binding activity of LBD was molybdate dependent, indicating a weaker association of mouse AhR LBD with Sf9 hsp90. A differential effect of Triton X-100 on the recombinant AhR LBD and native AhR also suggests that the interaction between AhR and Sf9 hsp90 is less stable. The study refined the minimal LBD to a region of 125 amino acids, which should be amenable for structural studies of the LBD.

The regulation of metabolic gene expression is fundamental to maintaining energy balance. Changes in substrate availability can alter metabolic gene expression, in order to modify the utilisation of nutrients appropriately. Metabolic gene expression can also be dysregulated in disease states. The work in this thesis examines several situations in which energy balance or metabolic substrate supply was altered, and investigates how metabolic gene expression adapted and contributed to the phenotypes observed following these interventions. All experiments in this work looked at metabolic regulation from a human perspective; with either biopsy material or cells cultured from biopsy. These experiments included; 1. The influence of postprandial fat-oxidising capacity during a calorie restricted diet of either high- or low-fat content in obese subjects. 2. Transcriptional profiling of adipose tissue in obese subjects with high- or low-postprandial fat oxidising capacity 3. High-glucose treatment of primary human myotubes (as a model of hyperglycaemia). 4. Increased PDC activation and hence carbohydrate oxidation in vivo, through administration of dichloroacetate. Postprandial fat-oxidising capacity did not affect weight-loss during a calorie restricted diet, and there was no affect of diet composition. However, changes in metabolic gene expression were observed between groups over the course of the 10-week intervention. The groups which showed the greatest changes in gene expression were the low fat-oxidisers on a high fat diet and the high fat-oxidisers on a low fat diet, possibly due to a mismatch between diets and fat oxidising capacity, which required greater adaptation. Covariate analysis revealed interactions between gene expression and other phenotypic parameters. SREBP-1c showed a relationship with FFA concentrations and insulin-resistance, whilst HSL and apM1 were associated with FFA concentrations and Insulin resistance respectively, which underlines the importance of looking for underlying structures in data. Transcriptional profiling of adipose tissue in obese subjects with high- or low-postprandial fat oxidising capacity, revealed significant differences in the expression of metabolic genes, and highlighted the importance of several transcripts; including RXRA, SREBP-1c and GLUT4 in determining the phenotype of adipose tissue. The major differences observed in gene expression between high and low fat oxidizers indicated that genes involved glucose metabolism and lipogenesis rather than beta-oxidation were the major processes that differed between the two groups. Genomic data indicated that the expression of these genes was not influenced to a major degree by polymorphisms within the population. High-glucose treatment of primary myotubes, demonstrated the significance of ChREBP and some of its targets, in inducing the expression of lipogenic enzymes, which may be linked to the accumulation of intramyocellular triglyceride. However, these data also indicated the potential for the cell to initiate protective mechanisms, of substrate handling and lipid clearance, in response to carbohydrate oversupply. Conversely, increasing PDC activity and hence carbohydrate oxidation without altering substrate availability via infusion of dichloroacetate, did not alter the expression of metabolic genes in skeletal muscle. This reflects a capacity to deal with acute changes in the activity in metabolic genes without altering their expression. In conclusion, the studies from this thesis show that important differences in metabolic gene expression can be observed during situations where energy balance and substrate availability are altered. However, flexibility within the metabolic networks means that acute changes can be countered without the need for induction or suppression of metabolic genes, and that during chronic alterations in nutrient supply, rapid adaptations and protective mechanisms are activated.

The intake of water and electrolytes is inseparable from feeding by natural or artificial means and careful attention to salt and water balance is a vital component of perioperative care and of nutritional support. Nutritional support with water and sodium restriction in post-intensive care patients with oedema, dilutional hypoalbuminaemia and fluid excess of 10 L, cleared oedema over 7-10 days, with a 1 g/L rise in serum albumin for every kg loss in weight. Return of gastrointestinal function was also observed. Accordingly, 20 patients, undergoing colonic surgery, were randomised to receive standard (>3 L water and 154 mmol sodium/day) or restricted postoperative fluids (<2 L water and 77 mmol sodium/day). Solid (72.5 vs 175 min) and liquid phase (73.5 vs 110 min) gastric emptying times were significantly longer in the standard group on the 4th postoperative day and associated with a three day longer hospital stay. In volunteers receiving 2 L of 0.9% saline and 5% dextrose infusions, on separate occasions over one hour, haematocrit and serum albumin concentration fell, mainly due to dilution. While dextrose was rapidly excreted, two-thirds of the saline was retained after 6 h. Following 1 L infusions, plasma renin and angiotensin concentrations decreased more after saline than dextrose (P<0.04). Responses of aldosterone, atrial natriuretic peptide and vasopressin were not significantly different. Comparing 2 L infusions of saline and Hartmann's solution, volunteers excreted more water (median 1000 vs 450 mL) and sodium (122 vs 73 mmol) after Hartmann's. Hyperchloraemia and reduced bicarbonate were noted after saline alone. Whereas fluctuations in water balance are dealt with efficiently through osmoreceptors and vasopressin, and sodium deficiency by volume receptors and the renin angiotensin aldosterone system, the mechanism for dealing with sodium and chloride excess appears relatively inefficient. Natriuretic peptide responds to volume expansion rather than sodium gain.

The pyruvate dehydrogenase complex (PDC) is the rate limiting step in the entry of glucose derived pyruvate into the tricarboxylic acid (TCA) cycle. As such it plays an important role in the control of the use of carbohydrate as the source of oxidative energy for skeletal muscle contraction. The first experimental chapter investigates the effect of dichloroacetate pre-treatment during low-intensity (<60% VO2max) exercise, below which it is suggested that increasing PDC activation and resting acetyl group availability via dichloroacetate (DCA) pre-treatment will be ineffective at reducing non-oxidative ATP production and improving contractile function. Despite a significant increase in both PDC activation (p<0.01) and acetylcarnitine availability (p<0.01) prior to the onset of exercise following DCA infusion, there was no difference in substrate level phosphorylation detected during exercise. The following experimental chapter examines the link between blood lactate concentration and the onset of the ventilatory threshold. Infusion of DCA (50mg.kg-1) prior to the onset of incremental exercise lead to a significant reduction in resting blood lactate (p<0.05), but this was not preserved during the following bout of incremental cycling exercise commencing at 50% VO2max. There was also no alteration in the onset of the ventilatory threshold detected after DCA pre-treatment. The final experimental chapter has investigated the effect of DCA infusion (50 mg.kg-1) upon high-fat diet induced PDC inhibition during exercise. During moderate cycling exercise (75% VO2max) DCA infusion reversed the high-fat induced inhibition of PDC activation. DCA infusion reduced the metabolic inertia present at the onset of contraction, through both PDC activation and pooling of acetyl groups prior to contraction. This thesis has highlighted the role of the PDC as an important site of regulation of human skeletal muscle fuel metabolism, which may provide a novel target for the treatment of the metabolic syndrome.

Pluripotent human embryonic stem cells (hESCs) are a valuable tool for clinical therapies, drug testing and investigation of developmental pathways. Recently, over-expression of four pluripotency-associated genes (OCT4, NANOG, SOX2, and LIN28) has proven sufficient to reprogram differentiated cells into pluripotent stem cells, potentially alleviating the need for human embryos to isolate hESCs and opening new avenues for the investigation of pluripotency. This project aimed to generate an in vitro model system to study the epigenetic mechanisms regulating pluripotency transcription factors. hESCs were differentiated into fibroblasts (hESC-Fib) and subsequently reprogrammed to induced pluripotency stem cells (iPSCs) by lentiviral over-expression of human OCT4, NANOG, SOX2 and LIN28. iPSC colonies were positively identified by live staining with the surface marker TRA-1-81 and expanded in culture. They were then further differentiated into a fibroblast line to allow comparison with hESC-Fib. All cells in the model system shared the same genotype and were cultured under similar conditions, enabling unbiased analysis of epigenetic characteristics. DNA methylation analysis of key pluripotency-genes such as OCT4, SOX2, NANOG, and REX1 by bisulfite sequencing, revealed that these were hypomethylated in hESCs and iPSCs, and hypermethylated in their fibroblast derivatives. A gradual increase in the number of CpGs gaining DNA methylation was observed when hESCs and iPSCs were differentiated into fibroblasts, while TaqMan real-time PCR and fluorescence staining revealed that expression of these genes was inversely related to the levels of DNA methylation in their promoters. The master pluripotency regulators OCT4, SOX2 and NANOG all showed differential methylation in their OCT/SOX binding regions, suggesting a common regulatory mechanism between them. This is, to our knowledge, the first report for SOX2 differential methylation in human non-cancerous cells. Reactivation of REX1 was not found to be necessary for the reprogramming of hESC-Fib to iPSCs, calling for re-evaluation of its role in human pluripotency. Based on the observation that the DNA methylation levels of pluripotency genes were higher in fibroblast cell lines compared to hESCs and iPSCs, we hypothesised that reduction in DNA methylation could render differentiated cells more permissive to reprogramming. Stable knock-downs of the DNA methyltransferases (DNMTs) DNMT1 and DNMT3A were, thus, performed in hESC-Fib. Knock-down of DNMT1, the most abundant DNMT in hESC-Fib, resulted in global reduction of DNA methylation levels as determined by restriction digests with methylation specific enzymes. Reprogramming of hESC-Fib carrying a DNMT1 knock-down showed a 40% reduction in generation of iPSC colonies compared to untreated controls, perhaps owing to the delay in progression of S phase in the cell cycle caused by DNMT1 knock-down. In contrast, knock-down of DNMT3A resulted in a >80% increase in iPSC colony formation, potentially indicating differences in mechanism of action and specificity between the two DNMT enzymes. Through this study, we have gained new insights into the epigenetic mechanisms underlying cell phenotype and provided the foundation for further improving reprogramming efficiency.

The covalent post-translational modification of selected substrates with the ubiquitin protein has emerged as a central regulatory mechanism, governing protein stability, activity and localisation, and accordingly an array of cellular processes. Ubiquitin signalling versatility arises owing to the diverse nature of (poly)ubiquitin modification, with distinct modifications subsequently transduced in a specific manner by ubiquitin-binding domains found in ubiquitin-binding proteins. In recent years the notion that ubiquitin exerts influence solely via the covalent modification of substrates has been challenged, with unanchored, or substrate-free polyubiquitin chains emerging as key regulators of cellular physiology. The investigations described in this thesis seek to exploit the inherent selectivity of natural and engineered ubiquitin-binding proteins, to afford the purification of endogenous unanchored polyubiquitin, probing the molecular composition and interactions of this biologically significant ubiquitin pool. In chapter 3 by utilising the deubiquitinating enzyme USP5, which contains multiple ubiquitin-binding domains, and is normally responsible for the selective disassembly of unanchored ubiquitin, we purify unanchored polyubiquitin from mammalian cell extracts. Subsequently, we apply both ubiquitin-selective antibodies and mass spectrometry-based analyses to examine the polyubiquitin profile of the mammalian unanchored ubiquitin pool. In chapter 4 we then assess the mechanisms of ubiquitin recognition by USP5, presenting a structural mass spectrometry-based framework to probe and quantify ubiquitin: ubiquitin-binding domain interactions. Finally in chapter 5, based on the conclusions we draw from USP5-ubiquitin recognition in chapter 4, that multiple domains in suitable arrangement yield specificity for polyubiquitin chains, we design and synthesize a synthetic protein to favour the capture of unanchored polyubiquitin chains of defined topology, namely Lys-48 linked diubiquitin (and longer polyubiquitin chains), from mammalian cell extracts. We conclude that strategies for the rational design and engineering of polyubiquitin chain-selective binding in non-biological polymers are possible, paving the way for the generation of reagents to probe the unanchored polyubiquitin chains of defined topology, and more widely the ‘ubiquitome’.

There are three prolyl hydroxylases (PHD1, 2 and 3) that regulate the hypoxia-inducible factors (HIFs), the master transcriptional regulators that respond to changes in intracellular O2 tension. In high O2 tension (normoxia) the PHDs hydroxylate HIFα subunits on 2 conserved proline residues inducing binding of the von-Hippel-Lindau (VHL) tumour suppressor, the recognition component of a multi-protein ubiquitin-ligase complex, initiating HIFα ubiquitylation and degradation by the 26S proteasome. However, it is not known whether PHDs and VHL act separately to exert their enzymatic activities on HIFα or as a multi-protein complex. In this thesis, data are presented that shows that the tumour suppressor protein LIMD1 acts as a molecular scaffold simultaneously binding the PHDs and pVHL into a normoxic protein complex (normoxiplex), increasing their physical proximity in order to enable efficient and rapid sequential modifications and thus degradation of HIF1α. Data are presented which indicates that increased LIMD1 expression down regulates HIF transcriptional activity, by promoting HIF1α degradation via the oxygen dependent degradation domain in a manner dependent on hydroxylase and 26S proteasome activities. However, degradation of this domain is not wholly dependent on the well characterised proline residues subject to hydroxylation, suggesting that LIMD1 may alter proline hydroxylation specificity or modulate HIF via a different mechanism. Furthermore, endogenous depletion of LIMD1 results in the converse, leading to HIF1α stabilisation and accumulation, enhancing HIF transcriptional activity. Moreover, Limd1-/- MEFs show increased HIF transcriptional activity. One mechanism by which this is achieved involves the binding of PHD2 within the N-terminal portion of LIMD1 while allowing concurrent binding of VHL to the C-terminal zinc-finger LIM domains. However, the LIMD1 mediated mechanism regulating HIF1α independently of proline residues 402 and 564 is still unclear. Finally, data are presented that show that the LIMD1 family member proteins Ajuba and WTIP all bind specifically to VHL but differentially to PHDs 1, 2 and 3 and thus these three LIM domain containing proteins represent a new group of hypoxic regulators.

書名據目次前. / 手稿覆寫本. / Thesis (M.A.)--香港中文大學. / Shu ming ju mu ci qian. / Shou gao fu xie ben. / Includes bibliographical references (leaves 302-307). / Thesis (M.A.)--Xianggang Zhong wen da xue. / Chapter 第一章 --- 導論 --- p.1 / Chapter 第二章 --- 六朝清商曲之淵源與發展 / Chapter 第一節 --- 清商釋名 --- p.13 / Chapter 第二節 --- 相和歌與三調曲之關係及運用之法 --- p.28 / Chapter 第三節 --- 六朝清商曲之運用與相和歌之關係 --- p.51 / Chapter 附 --- 子夜諸曲考 / Chapter 第三章 --- 六朝清商曲之興盛原因及衰亡 / Chapter 第一節 --- 六朝清商曲之流行 --- p.70 / Chapter 第二節 --- 六朝清商曲之興盛原因 --- p.84 / Chapter 第三節 --- 六朝清商曲之衰落 --- p.112 / Chapter 第四章 --- 六朝清商曲之內容 --- p.128 / Chapter 第一節 --- 吳聲歌與西曲歌 --- p.130 / Chapter 第二節 --- 神弦歌 --- p.168 / Chapter 第三節 --- 江南弄上雲樂與梁雅歌 --- p.181 / Chapter 第四節 --- 吳聲歌與西曲歌之匿別 --- p.194 / Chapter 第五章 --- 吳聲歌與西曲歌之特色及其表達方法 / Chapter 第一節 --- 吳聲西曲雙關諧聲詞格發達原因 --- p.207 / Chapter 第二節 --- 表達方法 --- p.222 / Chapter 第三節 --- 重複詞格 --- p.226 / Chapter 第四節 --- 女樂 --- p.229 / Chapter 第六章 --- 六朝清商曲對後世之影響 / Chapter 第一節 --- 對五言絶句之影響 --- p.234 / Chapter 第二節 --- 對詞之影響 --- p.253 / Chapter 附 --- 鐃選堂先生和聲與聯歌 / Chapter 第三節 --- 對唐法曲之影響 --- p.284 / Chapter 第七章 --- 結論 --- p.296 / 參考書目舉要

Background: Proton pump inhibitors (PPIs) have become the cornerstone of medical treatment for acid-related gastrointestinal disorders. To date, there is a distinct lack of understanding about the recent UK trends in PPI use and evidence about the association between the increased risk of these drugs and the potential adverse effects, in particular the risk of infection, remains questionable. The publication of contradictory findings in several research studies further compounds this situation. Aim and Objectives: This thesis aimed to examine the epidemiology of PPI use in general practices in the UK, and the side effects of PPI, mainly their proposed infective complications. The specific objectives were: • To determine the prevalence and pattern of PPI prescription, and to identify the practices employed to reduce PPI use in the UK general population. • To examine the risk of community-acquired pneumonia before and after the administration of PPI and to assess whether unmeasured confounding explains this association. • To determine whether the mechanism by which PPIs induce an increased risk of infection is supported by the same mechanism acting in another cause of achlorhydria, pernicious anaemia. Methods: This thesis describes work conducted using the UK’s Clinical Practice Research Database (CPRD) and, for some studies in this project, a subset of the CPRD linked to the hospital records from the Hospital Episodes Statistics (HES) database. Firstly, the CPRD was used to estimate the annual prevalence of PPI use during the period 1990-2013. In this study, new users of PPI therapy who had five years of follow-up data were identified to describe patterns of cessation and duration of PPI use. Secondly, cohort (analysed using Cox regression and prior event rate ratio) and self-controlled case series studies were conducted to examine the risk of community-acquired pneumonia and PPI exposure. Thirdly, a cohort of pernicious anaemia patients was used to estimate the risk of infections (community-acquired pneumonia and Clostridium difficile infection) compared to controls to examine whether a reduction in gastric acidity might be the underlying mechanism of the increased risk of these infections. Findings: 1) There was a considerable increase in the administration of PPI prescriptions in UK general practice such that both the period and point prevalence of PPI use increased between 1990 and 2012 (period prevalence increased from 0.2% to 14.8% and point prevalence from 0.03 % to 7.7%). Of new users of PPI therapy, 27% used PPI therapy over a long-term basis (≥1 year continuously), while 4% remained on PPI therapy for five years. Clear attempts to step down the dosage were identified in 41% of long-term users. 2) Among 320, 000 patients, including 160 ,000 new PPI users, the risk of community-acquired pneumonia was 1.67 (95% confidence interval (95%CI) 1.55 to 1.79) times higher for patients exposed to PPIs than it was for the controls. Among the 48,451 PPI-exposed patients with a record of community-acquired pneumonia, the relative incidence rate ratio was 1.19 (95%CI 1.14 to 1.25) in the 30 days after a PPI prescription but was higher in the 30 days before a PPI prescription (1.92, 95%CI 1.84 to 2.00). This reduction in the increased risk in PPI users after prescription was also reflected in the prior event rate of 0.91 (95%CI 0.83 to 0.99). 3) A total of 45,467 pernicious anaemia patients were identified and matched to 449,635 controls. Patients with a pernicious anaemia diagnosis had a higher risk of developing community-acquired pneumonia than the control group (adjusted hazard ratio(HR)1.24, 95%CI 1.21 to 1.26); however, this risk decreased when a stricter definition of pernicious anaemia was applied, and the data was further restricted to incident diagnosis. The findings also suggest that pernicious anaemia patients have a 57% increased risk of Clostridium difficile infection (adjusted HR 1.57, 95% CI 1.40 -1.76) and this association persisted when we limited the analysis to a subgroup with a more restrictive definition of pernicious anaemia diagnosis, or to incident cases. Conclusions: This research revealed that there was a high prevalence of PPI prescribing in the primary care setting and that there are considerable opportunities available to reduce the cost and side effects of PPI use through improving adherence to recommended withdrawal strategies. In addition, the studies investigating the proposed infective complications of PPI use on which we focussed in this thesis add important data to the development of a safety profile of PPI use.

Background and Aim: The cluster of differentiation 24 (CD24) is a human protein encoded by the CD24 gene which maps to chromosome 6q21. It is a small highly-glycosylated protein that is attached to the cell membrane via a glycosylphosphatidylinositol (GPI) anchor. Various studies have proven that CD24 is overexpressed in many human tumours, and its expression level has been found to be associated with a poor prognosis. Recently, CD24 has also been identified as a stem cell marker in several types of cancer. However little is known about CD24’s biological role in cancer or the mechanism through which it acts in cancer development. Thus, the main aim of this study is to investigate the potential functional role and signalling pathways of CD24 in various cancer models, such as colorectal cancer (CRC), pancreatic cancer, liver cancer, and lung cancer. Methods: The signalling pathway and functional role of CD24 has been studied in a total of 13 well-characterised cell lines from four cancer models i.e. cancers of colorectum, pancreas, liver and lung. To identify potential downstream targets of CD24, CD24 was forcibly expressed via transient transfection in cell lines expressing low levels of CD24 or, in contrast, it was knocked down using RNA interference (siRNA) in cell lines expressing high levels of CD24. Subsequent to CD24 manipulation, we used Western blotting and/or qPCR to investigate changes in the expression of specific proteins that have been hypothesised to be involved downstream in the CD24 signalling pathway e.g., C-terminal Tensin-like (Cten), focal adhesion kinase (FAK), integrin-linked kinase (ILK) and Src. In addition, changes in epithelial mesenchymal transition (EMT) markers were evaluated. Similarly, in order to find potential upstream regulators of CD24, the expression levels of some proteins that have been found to be associated with cancer (e.g., KRAS and EGFR) were manipulated using specific siRNAs or inhibitors/stimulators, and the changes in CD24 expression were then evaluated. The functional effect of these interventions were tested through measuring cell motility,invasion, proliferation, and stemness (by the colony-formation in agar). Physical interaction between proteins was tested by protein complex immunoprecipitation (CO-IP) and protein stability was tested using cycloheximide (CHX). Lastly, the subcellular localisation of CD24 was studied in CRC and lung cancer cell lines and, immunohistochemically, in tumour tissues of CRC (n=84) and NSCLC (n=58). Results: Our data have shown that manipulating of the expression of CD24 in the tested cancer model cell lines resulted in a significant change in the expression level of Cten which in turn caused changes in the expression levels of ILK and FAK. Noticeable modifications to cell migration, invasion, proliferation, and colony-forming rate (all p < 0.05) following CD24 manipulation have also been detected, indicating that the up-regulation of Cten, ILK and FAK expression by CD24 may reveal the mechanism via which cell functions are regulated. The up-regulation of Cten expression by CD24 was found to be due to protein stabilisation as confirmed by qPCR and CHX assy. CD24 was observed to activate AKT at Serine 473 (Ser473), rather than at the Threonine 308 (Thr308) residue, and potentially collaborate with PI3 kinase to induce the full activation of AKT. The inhibition of EGFR using a specific EGFR inhibitor, PD135053, and the stimulation of EGF using recombinant EGF in cell lines that did not harbour mutant KRAS resulted in significant modifications to CD24 expression, as well as in cell motility, suggesting that EGFR is an upstream regulator of CD24 expression. However, an inverse association between CD24 and KRAS was observed suggesting that EGFR does not signal to CD24 through KRAS. The signalling pathway would appear to be EGFR → CD24 →Cten →ILK/FAK→ AKT. These effects were seen in all of the models tested thereby confirming the role of CD24 in many cancers. In CRC and NSCLC cell lines and tissues, CD24 was found to be localised in both the cytoplasm and in the nucleus. An association between CD24 and its associated partners, including downstream targets, was observed in tumour tissues. This association was consistent with that observed following CD24 manipulation in cancer cell lines. Conclusion: CD24 seems to be regulated by EGFR either directly or indirectly. Consequently, it regulates Cten, FAK and ILK and enhances cell motility, invasion, proliferation and stemness in various cancer model cell lines. A combination of Anti-CD24 antibody/siRNA and PI3 kinase inhibitors could be used in clinical practice as a potential therapeutic agent in the early stages of CRC. Our observations in the four cancer models were consistent, suggesting that CD24 may regulate cell functions in these models through a unique mechanism. The association between nuclear CD24 expression and cancer progression and metastasis should be explored in further studies.