The high-density lipoprotein binding protein HDLBP is an unusual RNA-binding protein with multiple roles in cancer and disease

ABSTRACT

The high-density lipoprotein binding protein (HDLBP) is the human member of an evolutionarily conserved family of RNA-binding proteins, the vigilin protein family. These proteins are characterized by 14 or 15 RNA-interacting KH (heterologous nuclear ribonucleoprotein K homology) domains. While mainly present at the cytoplasmic face of the endoplasmic reticulum, HDLBP and its homologs are also found in the cytosol and nucleus. HDLBP is involved in various processes, including translation, chromosome segregation, cholesterol transport and carcinogenesis. Especially, its association with the latter two has attracted specific interest in the HDLBP’s molecular role. In this review, we give an overview of some of the functions of the protein as well as introduce its impact on different kinds of cancer, its connection to lipid metabolism and its role in viral infection. We also aim at addressing the possible use of HDLBP as a drug target or biomarker and discuss its future implications.

KEYWORDS:

• RNA-binding protein
• vigilin
• HDLBP
• biomarker
• hepatocellular carcinoma
• adenocarcinoma
• apolipoprotein B

The vigilin protein family

The human vigilin protein HDLBP (high-density lipoprotein binding protein) was first described in 1987 [1]. Originally identified as a supposed binding partner of high-density lipoprotein (HDL), it took until 1992 to recognize that the many repeats in the protein sequence are nucleic acid binding domains similar to those found in the heterologous nuclear RNA-binding protein (hnRNP) K Homology) domains [2,3]. The more commonly used term ‘vigilin’ for HDLBP and its homologs stems from a VIG (valine-isoleucine-glycine) motif situated at the N-terminus. Since it was not realized that the protein was already named vigilin became the common term for members of the whole protein family [4]. Vigilins are characterized by 14 KH (hnRNP K homology) RNA-binding domains taking up most of the protein sequence. In higher eukaryotes, a 15th KH domain (Figure 1) could later be identified in the N-terminus [5].

Figure 1. Schematic and predicted structure of vigilins. A) KH domain arrangement of two vigilins, the budding yeast Scp160 and the human HDLBP. Orange boxes represent classical KH domains, red ones represent diverged KH domains. B) COSMIC histogram indicating mutations across HDLBP in various cancer types. Mutations are depicted at an amino acid level correlating with HDLBP from a). Vertical lines display the location and relative frequency of substitutions C)Alphafold2 predicted protein structure of human HDLBP[6,7], using the same colouring scheme as in A).

Eukaryotic KH domains come in at least two different variations. The type I KH domains consist of a specific minimal β₁α₁α₂β₂-β₃α₃ motif with a GXXG loop between the two α helixes followed by an C terminal additional ß-sheet and an α-helix (Figure 2A). The type II domain consists of a α_A´β_a´β_aα_Aα_Bβ_b structure predominantly found in prokaryotes [8]. KH domains can bind several different nucleic acids via a hydrophobic groove formed by the specific pattern of the βααβ motif of the domain [9]. Vigilins, including HDLBP have been reported to bind to several types of nucleic acids, including DNA [10], tRNA [11], rRNA [12,13] and mRNA [8]. Notably, not all KH domains in the vigilin proteins are fully conserved. For HDLBP only 12 of its KH domains contain a classical, conserved, GXXG motif. The KH domains 0, 2 and 12 of HDLBP are so-called diverged KH domains, where the GXXG motif is either completely lost or altered (Figure 2B).

Figure 2. Architecture of a KH domain. A) definition of the minimal motif of a KH domain containing the GXXG loop between two α helices; B) comparison between classical (left) and diverged (right) KH domains in HDLBP (top) and the yeast vigilin Scp160 (bottom)[6,7].

Why does the protein contain so many KH domains? Firstly, individual KH domains can differ in their target RNAs [13]. It is thus conceivable that the protein contacts specific (sets of) RNAs via individual (sets of) KH domains. Furthermore, besides RNA, vigilins including HDLBP can also interact with proteins including those of the translational machinery. In affinity purification experiments HDLBP was found to be associated with translation elongation factors, tRNA and tRNA export factors [14]. Besides its interaction with tRNA and elongation factors cytosolic HDLBP co-fractionates with free and membrane-bound polyribosomes [15]. Further evidence for ribosome binding comes from the finding that the C-terminal KH domains 13 and 14 of the yeast vigilin Scp160p binds the small ribosomal subunit via the RACK1 homolog Asc1 [16]. The interaction of vigilin and RACK1 proteins was later confirmed for human RACK1 and HDLBP [17]. Based on above-mentioned findings, vigilins including HDLBP were suggested to act as RNA binders and participate in translation regulation. Despite early knockdown experiments of HDLBP showed no decrease in overall protein expression levels [18], a more recent study contradicted this. Zinnall et al. showed that HDLBP specifically binds to ~ 80% of mRNAs that are localized at the ER [13]. Knocking out HDLBP resulted in a decrease of translation especially of those mRNAs with a high affinity to HDLBP [13]. Consistent with their localization, most of these mRNAs encode membrane or secreted proteins or those of the endomembrane system. The expression profile of HDLBP coincides with these results since the protein shows high expression in tissue with high secretory activity like duodenum, pancreas or placenta (www.proteinatlas.org as of June 2023). Target secretory mRNAs are recognized by HDLBP via a seven-mer of variable UUC/UC/CUU/CU repeats in the coding region of the RNA [13,19]. This binding specificity is concomitant with a feature of the codons within the coding sequence of secretory mRNAs since CU is enriched in codons for hydrophobic amino acids such as Ile, Leu, Phe and Val [20]. Interestingly, vigilins do not only bind secretory but also cytosolic mRNAs. However, binding to the latter class occurs with lower affinity and mainly via the 3´UTR [13].

Besides its role in translation, vigilin proteins influence several physiological processes [21]. Among others, vigilin could be linked to control of p-body formation, stress granule composition, nuclear-cytoplasmic tRNA shuttling, gene imprinting and heterochromatin regulation, ploidy maintenance and sexual mating processes in Saccharomyces cerevisiae. For a more detailed description, we refer to a previous review on vigilin proteins [21].

Due to its multivalent functions, it is not unsurprising that HDLBP has turned up as a factor that modifies the severity of several diseases as well as a protein influencing cancerogenesis. HDLBP was found upregulated in several different cancers, can bind to viral RNA and is associated with atherosclerosis. However, the application of HDLBP as biomarker, diagnosis tool or therapeutic target is still in its infancy. In the following, we therefore summarize and discuss different studies on HDLBP for its medical usage.

HDLBP in lipoprotein metabolism

More than 30 years ago, it was discovered that vigilin is induced by the steroid hormone oestrogen in livers of Xenopus laevis [22] and in the epithelial uterus cells of ovariectomized rats [23]. Later, a similar – but inverse – regulation was observed by testosterone in testis of Xenopus laevis, where the hormone decreased the level of vigilin [24]. A connection between hormone-regulated vigilin expression and lipid metabolism was made by the observation that the mRNA of vitellogenin, an egg yolk precursor protein [25], is stabilized by vigilin via binding of the protein to vitellogenin 3´-UTR. Not only the expression of vigilin and vitellogenin but also binding of vigilin to vitellogenin mRNA is induced by oestrogen [26]. Vigilin was therefore promoted as an oestrogen inducible biomarker. However, oestrogen-dependent vigilin function and expression might not be a universal phenomenon since, for example, HeLa cells show no dependence on oestrogen for HDLBP expression [27]. These results indicate that vigilin is differentially regulated in different tissues or cell types [27], which coincides with the observation that vigilins in different organisms can also fulfill quite different roles [21]. Thus, transferring knowledge on vigilin function between different model systems should be interpreted carefully.

In addition to steroid hormones, the digestive hormone cholecystokinin octapeptide (CCK-8) can upregulate vigilin on RNA and protein levels [28]. In the pancreas, a major function of CCK-8 is the stimulation of expression and release of digestive enzymes like amylase and trypsinogen [28]. The simultaneous upregulation of digestive enzymes and HDLBP could be due to a co-regulatory effect but could also result from a promotion of vigilin expression simply due to an increased translation demand rather than upon specific hormonal activation [28]. In atherosclerosis plaques HDLBP was found especially expressed in foam cell macrophages and colocalized to the lipid transport protein ApoE (Apolipoprotein E) which is hypothesized to play a role in atherosclerosiss [29]. Their expression level can both be elevated in vitro by high cholesterol loading of cells. This indicates a similar area of effect for both proteins but also it could be shown that HDLBP expression is independent of ApoE. How these proteins interact in detail in the context of atherosclerotic lesions remains unclear [29]. A better-defined link between HDLBP function and development of atherosclerosis has been made for HDLBP via its binding to apolipoprotein B mRNA (ApoB) [19]. ApoB is another member of the lipid binding apolipoprotein family and a core component of low-density and very low-density lipoprotein particles (LDL and VLDL). These (very) low-density lipoproteins allow cholesterol and lipids to travel through polar solutions like the blood [30]. In order to investigate the role of HDLBP in lipoprotein metabolism in mouse liver, PAR-CLIP (photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation) experiments were combined with transcriptomic and proteomic analysis in HDLBP loss- or gain-of-function animal models [19]. A major finding was that around 17,5% of all mRNAs encoding proteins destined for the secretory pathway seem to be targeted by HDLBP. ApoB mRNA ranked as one of the strongest interaction partners [19].

Additionally, it could be shown that HDLBP binds mRNAs of ApoB, ApoC-III and and the glycoprotein fibronectin and that this binding depends on a CU-rich region [19]. Depletion of HDBPL via RNAi resulted in lower expression of two proatherogenic members of the apolipoprotein family (ApoB and ApoC-III) in the liver. Furthermore, the amounts of VLDL, plasma triglyceride and non-esterified fatty acids (NEFA) were lowered in the animals. In obese mice an elevation of ApoB correlated with an increased vigilin level. Healthy control mice, in contrast, did not show higher ApoB levels during vigilin overexpression. Therefore, higher lipid levels seem to a prerequisite for vigilin’s impact on apolipoprotein abundance.

In a more recent study, a liver specific knockout (LKO) mouse model for HDLBP in mice was generated. Similar to the RNAi knockdown model, 10 week old LKO mice showed downregulation of ApoB and ApoC-III, compared to their control littermates [31] HDLBP-LKO and control animals were comparable in bodyweight, fertility and plasma ALT (alanine aminotransferase) levels. However, a high-fat diet for six weeks decreased the plasma triglyceride and the cholesterol levels of the HDLBP-LKO mice. Silencing of vigilin in LDL receptor deficient (Ldlr⁻/⁻) mice, besides reducing the protein concentration of ApoB, lowered VLDL and LDL levels and decreased the formation of atherosclerotic plaques [19]. Since knockdown of vigilin did not affect the stability of ApoB mRNA, it is likely that vigilin acts as a translation efficiency enhancer, at least for ApoB. This indicates a to date only incompletely understood influence of HDLBP on the regulation of hepatic protein synthesis and lipid secretion via its translational impact [19].

HDLBP in cancer

Even today, cancer, with nearly 20 million deaths per year, still represents one of the leading causes of death worldwide. Research and treatment have made remarkable progress, but new approaches to detection and treatment of cancer need to be constantly explored. Over the last years HDLBP appeared in several studies as an upregulated protein in gastric cancer, ovarian cancer and prolactinoma, just to name a few (Table 1) [32–36].

Table 1. Summary of HDLBP-related cancer types and known interactors of HDLBP

Cancer type	Potential role of HDLBP	HDLBP interaction partner	Reference
Tuberous sclerosis complex (TSC)	exact function unknown	TSC2 (protein)	43,46
Malignant Mesothelioma (MM)	exact function unknown	unknown	55
Hepatocellular carcinoma (HCC)	driver gene; EMT signal modulation; suppressor of ferroptosis vulnerability	unknown	41, 42, 57, 61
Myxofibrosarcoma (MFS)	driver gene; promotes sarcomagenesis	unknown	37, 40
Leiomyosarcoma (LS)	driver gene; promotes sarcomagenesis	unknown	37
Osteosarcoma (OS)	driver gene; promotes sarcomagenesis	unknown	37
Small cell lung cancer (SCLC)	driver gene; promotes G1/S transition	unknown	53
Lung adenocarcinoma (LUAD)	driver gene	unknown	13
Breast cancer	tumor suppressor; inhibiting the expression of proto-oncogene	c-fms (RNA)	47, 48, 50

In a COSMIC (Catalog of Somatic Mutations in Cancer) dataset of frequently mutated genes in different cancer types, HDLBP could be identified as a notable example for altered genes in cancer patients. Over 1000 mutations within HDLBP from samples of cancer patients and human cancer cell lines were identified. When displaying all cancer-associated mutations found within HDLBP, no obvious hotspots are revealed (Figure 1B). However, for specific cancer types as myxofibrosarcoma [37] a more focused area of mutations can be identified. This indicates the necessity of assigning possible mutation hotspots to the corresponding cancer type. Homozygous deletions of the HDLBP gene were not only observed in sarcoma but also in cancer types like breast, lung, ovarian or epithelial cancer [38]. In myxofibrosarcoma and leiomyosarcoma, HDLBP is located in an analogous deletion peak. This was confirmed for other kinds of sarcoma [39]. Mutations detected in HDLBP in these cancers did not display a specific pattern in their distribution [38]. Concomitantly, upon induction of human sarcomas in mice those carrying a knockdown of HDLBP showed a boosted tumour development. In vitro knockdown of HDLBP in U2OS (osteosarcoma) cells led to an increase in soft agar colony formation. Due to these results, HDLBP has been suggested to act as a tumour suppressor [38].

In a screen for biomarkers to follow the response of soft tissue sarcoma to the kinase inhibitor pazopanib in patients, HDLBP was identified as a potential candidate [40]. Heterozygous or homozygous loss of HDLBP coincided with a decreased PFS (progression free survival; time in which a disease is not worsening) and treatment with pazopanib did not show an improvement of survival compared to tumours with both copies of HDLBP present. Consequently, the authors of this study suggested a role for HDLBP as tumour suppressor, too [40]. Deletions or mutations of the HDLBP gene have also been observed in myxofibrosarcoma [37]. Here, HDLBP was one of the more prominent cancer drivers most often altered in its copy number or mutated in a study of genetic variations in the landscape of myxofibrosarcoma [37]. In detail, two missense mutations and 22 biallelic deletions were identified in the HDLBP gene. Furthermore, these mutations of HDLBP were often coupled to aberrations in the tumour suppressor gene CDKN2A/B, suggesting that both genes act redundantly during the development of sarcoma. In addition, HDLBP alterations were found to accumulate in leiomyosarcoma with alterations in DSB repair genes like ATR, FANCA, BRCA1 and BRCA2. Taken together this suggests that HDLBP is involved in the sarcomagenesis in a yet unknown manner [37].

HDLBP misregulation has also been linked to several types of carcinomas, including hepatocellular carcinoma (HCC) [41]. Interestingly, in many of these cases HDLBP does not appear to act as tumour suppressor, but its expression correlates with tumour development. When comparing expression levels of HDLBP in HCC, healthy human liver, liver cirrhosis tissue and adjacent non-tumour liver tissue, high expression of HDLBP was only found in HCC tumour cells (29 out of 33 HCC samples) [41]. Since the levels of HDLBP increased from normal liver cells via cirrhosis and adjacent non-tumour hepatocytes to HCC cells, HDLBP could potentially be used as biomarker for HCC development. Comparison of three HCC cell lines and one embryonic hepatocyte cell line substantiated the theory connecting overexpression of HDLBP with development of HCC [41]. The protein level of HDLBP in all three HCC lines was way higher in comparison to the control embryonic cell line. To clarify the impact of HDLBP levels on HCC cell proliferation, a stable BEL7402 HDLBP-Knockdown cell line was generated, which showed a 33% reduction in cell proliferation versus the control line. BEL7402 is a hepatocellular carcinoma derived cell line. Survival, growth and proliferation of HCC cells therefore correlate with the protein level of HDLBP [41].

These findings match the results of recent studies where elevated HDLBP levels were found especially in forms of HCC with vascular invasion [42]. A lower three-year and five-year overall survival rate was directly correlated with higher HDLBP levels in HCC with vascular invasion compared to patients without, indicating a correlation of HDLBP expression with metastasis. This was corroborated by in vitro experiments showing impaired cell migration in hepatic Huh7 and Hep3B cell lines upon HDLBP knock-down and elevated migration when HDLBP was overexpressed [42]. Hints for the underlying mechanisms of these processes were obtained by GSEA (Gene set enrichment and western blot analysis) of HDLBP, which indicated that regulatory elements of the EMT (epithelial-mesenchymal transition) signalling pathway and mesenchymal markers were significantly increased while an epithelial marker was decreased [42]. The promotion of EMT by HDLBP depends on B-RAF, a proto-oncogene and important upstream kinase in the EMT signalling pathway.

Adenocarcinoma provides yet another example of HDLBP’s potential driving role in tumorigenesis. In a human A549 lung adenocarcinoma cell line, HDLBP deletion reduced the growth to 40–50% [13]. On the contrary, overexpression led to an elevated growth rate. Tumour formation experiments corroborated these results. While wild type A549 cells caused tumours in all injected mice, HDLBP deficient cells only caused palpable tumours in 37% of injected animals. These tumours were also decreased in size. Interestingly, mRNAs coding for secreted proteins (including those involved in extracellular matrix organization tissue development and cell adhesion) have been found to be downregulated in HDLBP KO strains [13]. This is consistent with the findings that the HDLBP primarily binds to (ER-) membrane localized mRNAs [13]. Loss of HDLBP leads to both downregulation (n = 700) and upregulation (n = 1039) of mRNAs, once more indicating the influential role of HDLBP in translation and turnover of transcripts important for tumour development [13]. Another way how HDLBP might impact tumour growth and proliferation is via its influence on stress granule formation/function [43]. A role for vigilin proteins in this process has already been described for the yeast homolog Scp160 [44]. Human stress granule components are elevated in tuberous sclerosis complex (TSC) derived tumours but how they impact tumorigenesis is largely unclear. TSC2 is a tumour suppressor gene, which, when mutated can cause TSC [43,45]. The TSC2 protein has been shown to bind HDLBP via the first six KH domains of the latter [46]. Both proteins accumulate in stress granules under heat and oxidative stress conditions and co-localize with the stress granule assembly factor G3BP1 [43]. Knockdown of HDLBP leads to a reduction of stress granules containing TSC2 but does not affect general stress granule formation, indicating a recruiting function of HDLBP for TSC2. The downregulation of the stress granule core component G3BP1 in TSC2 deficient cells in vivo resulted in an up to 75% decrease of subcutaneous tumour size. Taken together, the recruitment of TSC2 to stress granules by HDLBP and the decreased growth of TSC2 deficient cells after deletion of G3BP1 could be used to develop new therapies against TSC derived tumours [43].

For some target mRNAs, specific functional relations between HDLBP, the mRNA target and development of cancer have been described. One such example is the regulation of the proto-oncogene c-fms mRNA [47]. c-fms encodes the only cell surface receptor for the colony-stimulating factor (CSF-1) produced by macrophages [48]. c-fms overexpression enhances metastatic and invasive properties in breast cancer cells [49]. A 69-nucleotide element in the 3’-UTR of c-fms plays a central role in its regulation [50]. This element contains five ‘CUU’ triplets, a binding motif for HDLBP [13]. In the c-fms 3’-UTR, HDLBP competes for this element with HuR, another RNA-binding protein (RBP) known to target adenylate uridylate-rich elements [50,51]. Deletion analysis of the 69-nucleotide element in c-fms revealed that the full sequence is necessary for proper HDLBP binding and tumour suppression activity [50]. While HuR has a stabilizing effect on c-fms, HDLBP promotes mRNA decay and downregulation of c-fms translation, therefore acting as a repressor of c-fms expression and of breast cancer progression. This is one of the few reported examples where HDLBP acts as a tumour suppressor, opposing the general cancer promoting function apparent in this review [47].

An immunohistochemical approach was used to compare the expression of HDLBP in different breast cancer stages. Protein levels and localization of HDLBP were analysed in ADH (atypical ductal hyperplasia), DCIS (ductal carcinoma in situ), and IDC (Invasive ductal carcinoma). A decline of the protein abundance in comparison to normal breast tissue could be detected even in the early stage of ADH, which can pose as a higher risk of carcinoma development [52]. Another finding was a localization shift of the protein from the nuclear to the cytoplasm in stained DCIS tissue samples, underpinning the potential of HDLBP as a biomarker for cancer development [50].

Identifying new biomarkers is of utmost importance especially for early diagnosis as well as treatment of very aggressive cancer types like small cell lung carcinoma (SCLC). Due to the finding of HDLBPs as a possible biomarker for lung [53] and breast cancer [40], the identification of the protein in an aptamer screening against potential SCLC (Small cell lung cancer) markers, is of great interest. In this experimental setup using cell-SELEX [54] (systematic evolution of ligands by exponential enrichment) in combination with mass spectrometry HDLBP was found as a protein important for SCLC progression. In comparison to normal lung tissue HDLBP was positively expressed in 73% of all SCLC samples. The knockdown of HDLBP resulted in suppression of cancer cell migration, invasion and proliferation in vitro. In addition, it could be shown that suppression of proliferation was due to a modulation of the G1/S transition by HDLBP. Here, reduction of HDLBP led to a G1/S arrest, as indicated by a decreased cell count for S and G2/M phase cells and an increase in G1/S phase cells. To further investigate if HDLBP could be a possible target for therapy or a biomarker to follow tumorigenesis, its role was also investigated in vivo. Knockdown of the protein resulted in smaller tumours and reduced metastasis [53]. Similar findings could be made in other cancer types. Changes in the amount for HDLBP could be identified in malignant mesothelioma (MM). The classification of MM is problematic since it is challenging to differentiate it from LAC (lung adenocarcinoma), ovarian cancer, or pleurisy, to just name a few examples [55]. To distinguish between MM types, an immunohistochemistry assay for the presence of HDLBP was performed. It could be shown that epithelioid, sarcomatoid and biphasic MM, as well as LAC (lung adenocarcinoma) and LSC (lung squamous cell carcinoma) showed different abundance of HDLBP, indicating different expression levels of this protein. Therefore, in immunostaining diagnostics HDLBP has potential to be used for the differentiation between MM and LAC [55]. Taken together, HDLBP is clearly linked to cancer progression. The molecular or functional connection is likely directly related to the reported functions of the HDLBP as vigilin proteins are involved in the synthesis of transmembrane proteins as well as proteins in the secretory pathway. For example, mice HDLBP was found to bind CU-rich areas of mRNA encoding primarily secretory proteins [19]. Similar target mRNAs have been identified for yeast Scp160 [56] and human HDLBP mRNAs [13]. Although, these targets includewell-known drivers of cancer progression including the transferrin receptor 1 (TfrC) [57]. These findings indicate that HDLBP by its very nature of binding and controlling multiple mRNAs coding for secreted and transmembrane proteins can, instead of a specific role, also have a rather promiscuous and indirect impact on cancer progression.

HDLBP and its unexpected role for at the crossroads of sterol metabolism and cancer

HDLBPs role in regulating apolipoprotein production as well as its binding to mRNAs encoding proteins of the transmembrane system suggests a connection of this RBP to lipid metabolism. In fact, one of the first characterization attempts to elucidate the function for HDLBP suggested a role in regulation of sterol metabolism [3] since exposure of cells to high cholesterol resulted in overexpression of the HDLBP, suggesting a protective function towards an accumulation of intracellular cholesterol [3].

Deregulated lipid metabolism can increase the risk of hepatocellular carcinoma (HCC) [58], impair the efficiency of potential anti-cancer treatment [42], and lower the overall survival rate of HCC patients [42]. Therefore, interfering with HDLBP function could be of great promise for cancer therapy, especially since it has been shown to be a modulator of potential anti-cancer drugs. A connection between HDLBP and such drugs has been demonstrated for simvastatin, a cholesterol synthesis inhibitor [42]. Treatment of HCC cells with simvastatin resulted in a downregulation of HDLBP expression as well as lowered migration and invasion in vitro. This effect could be countered by overexpressing HDLBP, which restored EMT, as well as invasive and migration behaviour of the cells. Cholesterol promoted the binding of HDLBP to B-RAF (a proto-oncogene and kinase in the EMT signalling pathway), while simvastatin had an inhibitory effect [42]. Furthermore, the half-life of B-RAF was decreased upon knockdown of HDLBP, while its overexpression increased its half-life. Interestingly, the role of HDLBP in this process appears to be independent of its function in mRNA homoeostasis. The regulatory effect arises from the competition of HDLBP with the Itchy E3 Ubiquitin Protein Ligase (ITCH) for B-RAF binding [42]. ITCH ubiquitinates B-RAF, thus promoting its degradation [42,59]. This kind of regulation by HDLBP is also seen for other members of the RAF family [59]. RAF1 (or C-RAF) was found as a direct interactor of HDLBP, and while overexpression of HDLBP did not elevate RAF1 mRNA levels, it raised the amount of RAF1 protein via interfering with the ubiquitin-proteasome degradation pathway [59].

Since RAF1 is one of the main targets of the therapeutic protein kinase inhibitor sorafenib, an intriguing question is that of a connection between sorafenib resistance observed in cells overexpressing HDLBP and the complex formation of RAF1 and HDLBP. HDLBP promotes RAF1 expression and regulates the kinase activity of MEKK1 towards serine 259 of RAF1, thereby activating the RAF1-MAPK signalling pathway [59]. Concurrently, HDLBP competes with the E3 ubiquitin-protein ligase TRIM71 for binding to RAF1, thereby inhibiting the degradation of the protein. Both mechanisms contribute to the stabilization and activation of RAF1, resulting in HCC growth, proliferation and sorafenib resistance [59]. On the contrary, HDLBP can also enhance the effects of cancer treatment as indicated by the observed interaction of HDLVBP and aralin, a type II ribosome-inactivating protein from the plant Aralia elata. Aralin showed promising effects on HDLBP deregulated tumours and has a selective cytotoxic effect on different cancer types including leukaemia, cervical, liver and lung cancer [60,61]. HDLBP was identified as an aralin binding partner and supposedly mediates aralin’s access to the cell via receptor-based endocytosis. Interestingly, the interaction of aralin was limited to a processed 110kDa version of HDLBP and was correlated with the cholesterol levels of the cells [62].

Another connection of HDLBP to cancer treatment is its involvement in ferroptosis. Ferroptosis is a type of cell death dependent on iron and, due to its non-apoptotic characteristics has been suggested as of potential use for cancer treatment [63]. Therapeutic agents like Lenvatinib and sorafenib can be used to trigger ferroptosis in cancer cells. The mRNA levels of the cyclooxygenase PTGS2 (prostaglandin-endoperoxide synthase), a marker of ferroptosis, are dependent on HDLBP. Elevated levels of HDLBP decreased the amount of PTGS2 mRNA while lower levels led to an increase [63]. Longer addition of sorafenib to a HCC cell line resulted in an increase of HDLBP levels, indicating HDBLP to thwart the vulnerability to ferroptosis. Overall, an overexpression of HDLBP resulted in a decrease of efficiency of ferroptosis-inducing reagents in HCC. How HDLBP counters ferroptosis is largely unknown. A hint comes from the observation that it stabilizes the large non-coding RNA lncFAL, which itself restricts effects of ferroptosis-inducing reagents. lncFAL expression results in a reduced degradation of FSP1, a ferroptosis suppressor protein [63].

HDLBP and viral replication

Like several other cellular RBPs, HDLBP has been implicated in viral RNA replication after infection of host cells. In particular, this case has been made for single-stranded plus-sense RNA viruses, including flaviviridae, caliciviridae or SARS-CoV19 [64–67] (Figure 3). Due to HDLBP’s shaping role for the membrane and secreted proteome, it is likely that loss of HDLBP affects infection by many viruses, e.g. via interfering with the levels of viral receptors on the plasma membrane or via affecting production of viral membrane proteins [68]. However, not all viruses are dependent on HDLBP, since specific groups like flaviviridae (see below) but not members of the alphavirus group are directly affected by loss of HDLBP [65].

Figure 3. Simplified representation of a single-stranded positive-sense RNA virus infection; interaction of viral RNA (red) with HDLBP (green) during its translation at the ER. At the ribosome, HDLBP is bound to RACK1 (yellow) in proximity of RRBP1 (purple). Different compartments and steps during viral infection are indicated. I. plasma membrane; II. endoplasmic reticulum; III. nucleus; IV. Golgi apparatus. 1. Viral entry and disassembly; 2. Viral replication, including 2a. translation, 2b. transcription, 2c. viral RNA replication, and 2d. virion assembly.

One of the first viral RNAs discovered to be bound by HDLBP was that of norovirus [66]. Norovirus, of the family of caliciviridae, is the most frequent cause for gastroenteritis in humans [69]. It could be shown that HDLBP binds the 5´ end of the viral RNA. Direct RNA-binding to several flaviviridae family members has also been shown for HDLBP. Flaviviridae are small enveloped viruses that infect birds and mammals and this group includes many important human pathogens like hepacivirus, yellow fever virus and the dengue virus [70]. In an RNA-centric approach aimed at identifying RBPs that bind to specific viral RNAs (ChIRP-MS; comprehensive identification of RBPs by mass spectrometry), HDLBP was identified as a yet unknown host RBP [65]. In this study, three flaviviral RNAs were tested for their interaction with HDLBP, those from Dengue virus (DENV), Zika virus (ZIKV) and Powassan virus (POWV). Binding studies were complemented by loss of function analysis, which revealed that the loss of HDLBP reduces viral RNA and protein levels, indicative of a role as a host protein in viral RNA replication and translation. To better understand the binding of HDLBP to viral RNAs and to identify co-binding partner proteins, a RNase A treatment was performed. In untreated cellular extracts HDLBP co-precipitates with RRBP1, a ribosome-binding protein of the endoplasmic reticulum membrane [71]. This interaction is lost upon RNAse A treatment, suggesting that the interaction of the two proteins is mediated by RNA. Upon DENV and ZIKV infection, both HDLBP and RRBP1 bind at the ER in a viral RNA dependent manner. Concomitantly, binding to host mRNA is lost. irCLIP (infrared-crosslinking immunoprecipitation) experiments [72] were performed to obtain a more detailed view on the binding profile of HDLBP to RNAs upon viral infection. 75% of all observed HDLBP-RNA crosslinks in DENV infected cells the contacts were of viral RNA origin. Upon zikavirus infection, 49% of all RNA crosslinks occurred to viral RNA, preferably in exonic regions, demonstrating a massive recruitment of HDLBP to viral RNAs. Further studies showed that the role of HDLBP in viral infection is more important in the later stages, and that it helps promoting translation of the viral RNA. This is consistent with the location of HDLBP and RRBP1 at the ER and the observation that replication of zikaviral RNA and virion assembly, which also requires translated viral proteins both occur at the ER [73]. How HDLBP affects viral biogenesis has yet to be shown. However, as Zinnall et al. [13] suggest, HDLBP might remodel the ribosome’s elongation arrest around the targeting signals, which is required for a more efficient nascent peptide targeting and translocation into the ER lumen. In addition, aggregation, degradation and misfolding could be reduced by peptide elongation control. HDLBP might also operate in this manner during flaviviral replication and synthesis of virions.

While further investigating HDLBPs role in flavivirus infection, RACK1, a component of the 40S small subunit of the ribosome came into the focus. In yeast, the corresponding homologous protein named Asc1p binds directly to the yeast homolog of HDLBP (Scp160p) [16], and these two conserved proteins bind each other in human cells as well [64]. Knockout of RACK1 results in a decreased Dengue virus infection rate in HAP1 cells [64]. To see if binding partners of RACK1 also affect the viral infection, the 49 highest ranking interaction partners of RACK1 were silenced by RNAi during DENV infection. HDLBP again was one of the hits which drew the interest of the investigators, due to the decrease of viral infection by 50% upon HDLBP depletion. It was thus hypothesized that the viral infection promoting effect of HDLBP could be dependent on its binding to RACK1. To test this, the last two KH domains of HDLBP were deleted as these are responsible for interaction of Scp160p with Asc1p [16,63]. As expected, the interaction of HDLBP and RACK1 as measured by co-immunoprecipitation was abolished when KH domains 13 and 14 of HDLBP were absent. In contrast, the binding of viral RNA to the truncated protein was not impaired, unlike the situation in yeast where loss of KH13 and KH14 reduces mRNA binding [74]. Despite RNA binding remained unchanged, the truncated HDLBP could no longer support viral replication of DENV, which strengthens the argument that binding to RACK1 is critical for HDLBPs role in flaviviral replication [64]. Targeting this interaction by small molecules might lead to new strategies to fight flavivirus infections.

Outlook

While many of its molecular functions may still be not be understood in detail, HDLBP could present a new entry point for biomedical applications, including its use as specific biomarker as shown for lung and breast cancer. It might also become a promising marker to discriminate between otherwise hard to distinguish cancer types like malignant mesothelioma and lung adenocarcinoma. Although its molecular impact on tumorigenesis or cancer propagation is not clear for most cases, for some cancer types including malignant mesothelioma or hepatocellular carcinoma a promoting effect by HDLBP appears likely. In other cases, such as breast cancer, HDLBP might serve as a tumour suppressor gene. Since HDLBP has been established as an RBP with 15 KH domains, these roles might be mediated by influencing yet unknown target RNAs. However, binding of specific protein partners like the tumour suppressor TSC2 also occurs via an N-terminal region containing six KH domains. Similarly, the C-terminal two KH domains are important for the interaction with the ribosome-associated factor RACK1. Thus, targeting specific regions or domains without impairing the overall functionality of HDLBP by mutating or deleting specific KH domains could turn out as a new promising approach to alter specific HDLBP functions.

Acknowledgments

We would like to thank L. Heinold for critical reading of the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data sharing not applicable – no new data generated.

Additional information

Funding

The work was supported by the Deutsche Forschungsgemeinschaft [JA696/10-2].