Post-translational modifications of the linker histone variants and their association with cell mechanisms

Authors


C. M. Wood, School of Pharmacy and Biomolecular Sciences, Liverpool John Moores University, Liverpool, UK
Fax: +44 0 51 298 2624
Tel: +44 0 51 231 2565
E-mail: c.m.wood@ljmu.ac.uk

Abstract

In recent years, a considerable amount of research has been focused on establishing the epigenetic mechanisms associated with DNA and the core histones. This effort is driven by the fact that epigenetics is intimately involved with genomics in a whole range of molecular processes. However, there is now a consensus that the epigenetics of the linker histones are just as important. The result of that consensus is that the post-translational modifications (PTMs) for most of the linker histone variants in human and mouse have now been established by a number of experimental techniques, foremost of which is mass spectrometry (MS). MS was also used by our group to establish the PTMs of the linker histone variants in chicken erythrocytes. Although it is now known which types of PTM occur at particular locations on the linker histone variants, there is still a large gap in the knowledge of how this data relates to function. The focus of this review is an analysis of the PTM data for the linker histones from several species, but with an emphasis on human, mouse, and chicken. Our analysis reveals that certain PTMs can be clearly correlated with specific functions of the linker histones in particular cell types, and that unique PTM patterns exist for different cell types.

Abbreviations
CDK

cyclin-dependent kinase

DNMT

DNA methyltransferase

HDAC

histone deacetylase

miRNA

microRNA

MS

mass spectrometry

PTM

post-translational modification

Introduction – epigenetic mechanisms and involvement with disease

Epigenetics is the study of heritable changes in gene expression that occur without changes in DNA sequence and, as well as being of fundamental importance in embryonic development, transcription, chromatin structure, X-chromosome inactivation, and genomic imprinting, it is also now recognized as having a fundamental role in disease [1]. RNA silencing, DNA methylation and post-translational modifications (PTMs) of the core and linker histones are the mechanisms that collectively define epigenetics, the latter of which involve the addition of small chemical groups. The PTMs that are created by this mechanism include, but are not limited to, acetylation (lysine), phosphorylation (serine, threonine), methylation (lysine, arginine), sumoylation (lysine), and ubiquitination (lysine). Other epigenetic mechanisms may emerge in the future.

Small RNAs

MicroRNAs (miRNAs) are RNA molecules that are about 22 nucleotides long and encoded into the Homo sapiens (hereafter ‘human’) genome [2]. They are transcribed by RNA polymerase II into primary miRNAs and afterwards processed by RNase III Drosha and DGCR8 in the nucleus into precursor miRNAs. These precursor miRNAs are then exported by Exportin-5 to the cytoplasm, where they are further processed by RNase III Dicer into the mature miRNAs [2]. Each miRNA is thought to have many targets and can bind its target mRNA completely or partially. If there is complete binding, the mRNA is silenced and degraded; partial binding leads to downregulation of a gene. It is known that miRNAs are related to small interfering RNAs and have similar functions. As small interfering RNAs have been shown to be involved with DNA methylation and histone modifications, it is likely miRNAs operate in the same manner [2]. The fact that miRNAs are located within the introns of protein-coding genes has led to the belief that they are activated with their host genes. A potential way that this may be achieved is via an active chromatin hub [3].

DNA methylation

In humans, DNA methylation occurs at a cytosine that precedes a guanine in a CpG dinucleotide sequence, most often occurring in short stretches of CpG-rich regions known as CpG islands. Such regions are about 0.5–2 kb long and can be found in the 5′-region of approximately 60% of genes, near to their promoters [4,5]. The cytosine base is modified at the 5-carbon position of the pyrimidine ring by the covalent addition of a methyl group (CH3) [5]. This modification is mediated by DNA methyltransferases (DNMTs) acting in concert with S-adenosylmethionine, which acts as a methyl donor in the enzymatic reaction. It is believed that the pattern of DNA methylation is established in germline cells through the action of de novo DNMT 3a and DNMT 3b. This pattern of DNA methylation is maintained subsequent to DNA replication through the action of DNMT1. The linker DNA can be preferentially methylated in the absence of H1, but the presence of the latter will inhibit methylation [6]. CpG dinucleotides are uniformly dispersed in humans, probably because 5-methylcytosine can be spontaneously deaminated to form the DNA base thymidine. CpG dinucleotides outside islands are essentially continuously methylated, leading to the genes where they reside being unexpressed. This is a necessary feature, as there is a large amount of noncoding DNA in the human genome. However, within CpG islands, the dinucleotides can be either unmethylated, if the gene is expressed, or methylated, if it is not expressed. There are two exceptions to this rule: imprinted genes, and genes associated with X-chromosome inactivation will always have their CpG islands methylated.

Histone modifications

The N-terminal tails of the core histones extend beyond the nuclesomes and can have their characteristics significantly altered by PTMs. H3 has the greatest number of modifications currently identified, followed by H4, H2B, and H2A. The C-terminal tails also contain PTMs, but they are few in number, as are those for the non-tail regions. Lysine acetylation weakens electrostatic DNA–histone interations, allowing the recruitment of factors containing bromodomains such as SWI/SNF and TFIID [5]. Methylation of H3 Lys10, H3 Lys28 and H4 Lys21 has been associated with gene silencing, whereas H3 Lys5, H3 Lys37 and H3 Lys80 (genomic position numbering) correlate with actively transcribed genes. It is not only the core histones that are subject to PTMs; the linker histone H1 can also be modified (see later).

Epigenetic mechanisms in disease

As specific pathologies (syndromes) can be associated with problems in the epigenetic machinery, and epigenetics is fundamental to chromatin structure, those diseases have become generically known as diseases of chromatin. For example, abnormal DNA methylation can cause errors in genomic imprinting, with an increased risk of Angelmann’s syndrome [7]. However, epigenetic problems are also implicated in many more frequently-occurring diseases, such as cancer.

Many cancer types have been shown to have gains in methylation at CpG islands in the promoters of some key genes. Such modifications are associated with transcriptional inactivation [8]. The gains in DNA methylation, or hypermethylation, are responsible for the underexpression of tumour suppressors such as p16INK4a and BRCA1 [5]. Early methylation of DNA may be a sign of tumorigenesis, as happens to the Wnt pathway in colon cancer [9], and DNMTs are often overexpressed in solid and wet cancer types [10]. Mutations and amplification of the androgen receptor gene, without loss of gene expression, play a key role in the development of advanced, androgen-independent prostate cancer [4]. Methylation of the androgen receptor promoter is prevalent in androgen-independent prostate cancer, but less so in androgen-dependent prostate cancer [4]. As well as hormonal genes, cell cycle genes are also affected in prostate cancer; an example is the methylation-mediated inactivation of the CDKN2A gene [4]. Methylation also goes awry in haematopoietic malignancies, and hypermethylation of p16INK4a has been observed in non-Hodgkin’s lymphoma, multiple myeloma, and acute lymphocytic leukaemia [8].

It is widely accepted that DNA methylation should, in the right circumstances, be a target for clinical treatment. Accordingly, nucleoside inhibitors that inhibit DNA methylation, such as azacitidine, decitabine, and zebularine, have been developed. All three are cytidine derivatives that irreversibly inhibit DNMTs [11]. As decitabine contains a deoxyribose group, it is incorporated into DNA [12]. However, because azacitidine contains a ribose group, it is initially incorporated into RNA [12]. Incorporation into DNA occurs when azacitidine is converted into 5-aza-2′-deoxycytidine diphosphate by ribonucleotide reductase, which is then phosphorylated, the triphosphate form being incorporated into DNA in place of the natural base cytosine [12]. The use of such analogues results in the global depletion of DNMTs and a subsequent reduction in DNA methylation.

Although DNA methylation is the most studied epigenetic modification in terms of clinical diagnostics, the mechanism is also important for histone modifications. DNMTs can interact with histones in two ways. First, DNA methylated by DNMT can attract proteins such as MeCP2 that are able to recruit histone deacetylases (HDACs); and, second, DNMTs can themselves directly recruit HDACs to help silence gene expression [4]. Most of the literature on interactions with methylated DNA has centred on the core histones H2A, H2B, H3, and H4, but a complete picture of epigenetic modifications cannot be obtained until linker histone PTMs have been factored in. This review analyses the research effort expended thus far on linker histone PTMs. For consistency, amino acid positions in a sequence are referred to by their actual genomic position, as given in Swiss-Prot and similar databases.

Structure and function of the linker histone variants

Location and structure of the linker histones

Historically, the location of the globular domain of the linker histone has been a matter of contention [13]. Currently, although there is a good degree of agreement about the overall parameters of the fibre formed by folding the zig-zagging chain of nucleosomes in inactive chromatin, the location of the linker histone in relation to the nucleosome core particle and linker DNA is still not known to high resolution (Fig. 1). However, recent studies [14–16] suggest that the linker histone is close to the dyad axis of the core particle at the entry and exit of the DNA. Similarly, the geometry of one nucleosome in the fibre relative to a DNA-connected nucleosome is also unknown [17–20].

Figure 1.

 The possible locations of the linker histone in relation to the nucleosome core particle. The globular domain of the linker histone will be located either symmetrically (left image), or asymmetrically (right image) [14,15]. Colour assignments are as follows: magenta, nucleosome core particle; blue, 146 bp of DNA; red, globular domain of linker histone H1; green, 22 bp of linker DNA; orange, 11 bp of linker DNA.

The structure of the linker histone H1 in humans is characterized by a relatively short N-terminal tail, a longer C-terminal tail, and a conserved globular domain [21]. This model extends to most other organisms, two exceptions being Tetrahymena thermophila and Saccharomyces cerevisiae [22]. The linker histone H1 variants show the greatest diversity when compared to the core histones. There is also a great diversification in the H1 variants within a single species such as H. sapiens, predominantly in the N-terminal and C-terminal tails, with, as stated, a conserved globular domain. However, when similar H1 variants are compared between species, there is a remarkable similarity. The nearer the species, the less is the divergence, such that H1.2 in Pan troglodytes has just one amino acid difference from its human counterpart, and the H1.4 variant in the human and Mus musculus (hereafter ‘mouse’) genomes has 93.6% sequence identity (Fig. 2). The reason for this is that the H1-variant genes within a species are paralogues, originating from gene duplication events, whereas the same H1 gene between species is an orthologue, originating from an ancestral gene [23]. In humans, the variants consist of the following: the somatic subtypes, H1.1–H1.5; a spermatogenesis subtype, H1t; an oocyte-specific subtype, H1oo; and a replacement subtype, H1o. H1.1–H1.5, along with H1t, are known as the replication-independent group, and are mainly expressed in S-phase. The remaining two, H1oo and H1o, are known as the replication-dependent variants. H1.1–H1.5 and H1t reside on the short arm of chromosome 6; H1oo is located on chromosome 3 and H1o on chromosome 22. H1.2 and H1.4 predominate in most cell types. The affinity of the various types seems to depend on their C-terminal tails. H1.1 and H1.2, with the shortest C-terminal tails, a low density of positively charged residues, and the lowest number of cyclin-dependent kinase (CDK) sites, have the lowest affinity for chromatin. A CDK site is identified by the consensus sequence (S/T)PXZ, where X is any amino acid and Z is a basic amino acid. H1.4 and H1.5, with longer C-terminal tails and more than two (S/T)PXZ sites, have the highest affinity for chromatin. With the highest content of positively charged residues, H1.3 has an intermediate affinity for chromatin [23]. The precise functions of the H1 variants within a cell are only just starting to be elucidated.

Figure 2.

 Phylogeny tree of human and mouse linker histones. Speciation events are indicated by blue dots and gene duplication by red dots. HIST1H1C and HIST1H1E are the genes that code for human linker histones H1.2 and H1.4, respectively. Note how the human HIST1H1C and HIST1H1E genes have a last common ancestor that is a duplication node, which makes these two genes paralogues. However, HIST1H1C in human and mouse originate from a speciation node, and are therefore orthologues. The phylogeny tree was generated by treefam [69].

Functions of the linker histones

It was evident from work with unicellular organisms that the linker histones were not critical for growth and cell division [23–25]. Following these experiments, it was speculated that the H1 variants were not global repressors of transcription, and this has now been shown to be the case [25]. Depletion of H1 in mammals causes significant changes to chromatin structure. When chromatin is depleted of H1, there is a reduction in the nucleosome-repeat length globally and a reduction in local chromatin compaction [25]. The reduction in repeat length arises from having fewer than one linker histone per nucleosome [25]. Depletion of H1 in mammals also causes a reduction in H3 Lys28 acetylation, with a smaller reduction in H3 Lys28 trimethylation, and also leads to a reduction of methylation at CpG islands in some of the H1-regulated genes [25]. The H1 variants tend to associate with specific transcriptional regulators [23]. For example, H1.1 specifically associates with BAF, which regulates chromatin structure [23], and H1.2 has been shown to associate with p53 [26]. It is thought that the specificity of individual variants stems partially from their sequence diversity, but mostly from PTMs [23]. Thus, the evidence emerging is that the H1 variants have specific functions. First, individual H1-variant knockout mice gave rise to specific phenotypes, with distinct effects on gene expression and chromatin structure [27]. Second, in the knockout mice referred to, there was no equal upregulation of the remaining variants, with only particular variants being able to compensate. Third, there are differences in the localization of the H1 variants within the nucleus, and there are variations in their relative amounts between different cell types [28]. Fourth, the H1 variants have different affinities for chromatin, can be recruited to specific transcription factors, and, as we shall see, have particular PTM patterns.

Significantly, H1.2 is now associated with an extranuclear function. This variant will, upon a DNA double-strand break, translocate to the cytoplasm and then permeabilize the mitochondrial membrane, causing the release of apoptotic compounds [29–31]. It has been shown that H1.2 in a cell infected with a virus displays an increase in mobility [32], and this property may play an important role in the treatment of cancer [33].

A specific mutation in H1.4 was detected in Raji cells, but was not detected in 103 healthy individuals or other Burkitt’s lymphoma cell lines [34]. This could be an example of H1 sequence variation acting as a marker for a particular phenotype. However, the main aim of Sarg et al. [34] was to demonstrate the use of a particular chromatography technique, rather than to firmly establish H1 sequence variation with disease. Thus far, then, no in-depth analysis has been performed that attempts to correlate linker histone sequence or PTM variation with disease. This is not surprising, especially in the case of the latter, as there is a potential for many permutations. Nevertheless, this must be the next phase in the work on H1 PTMs.

H1o is a general differentiation-dependent linker histone, and has a similar sequence to the avian H5 variant. H1o will accumulate in a cell, reaching a peak at terminal differentiation, being initially synthesized in oocytes and early embryos [35].

Epigenetic control of linker histones was discovered relatively early, with studies of synchronously dividing nuclei in the plasmodia of Physarum polycephalum [36], when it was shown that phosphorylation of linker histones is strongly implicated in cell cycle control and that phosphorylation is a precursor of mitosis. It is now widely accepted that patterns of PTMs on the core histones can influence transcriptional activity. For example, acetylation of H3 Lys10 has an inverse relationship with the amount of DNA methylation [37,38]. As it is also known that the linker histones can affect DNA methylation [25,39], it is not unreasonable to conclude that there must be key PTM patterns that govern – or at least significantly contribute to – the function of H1, as they do in the core histones. Although much work has been done to identify specific PTMs in the linker histones, there is still a gap in our knowledge of how these affect function.

Phosphorylation of CDK and non-CDK consensus sites

A PTM pattern for mitosis?

Early work on identifying the sites of phosphorylation in core and linker histones indicated that there was no correlation between cell cycle status and the number or location of phosphorylation sites [40]. Later work, however, has shown that this is not the case [41], and that the H1 linker histone of human lymphoblastic T-cells has phosphorylation states that correlate with the interphase and mitosis stages of the cell cycle. During interphase, it was found that the H1.5 variant was phosphorylated at Ser18, Ser173 and Ser189, which all reside in a CDK consensus motif of the form (S/T)PXZ, as previously defined. It was found that, during mitosis, the same three serine phosphorylations were present, but were also accompanied by phosphorylations at Thr11, Thr138, and Thr155. The first of these three threonines is not located within a TPXZ consensus sequence, but the latter two are. The same pattern of cell cycle dependency of phosphorylations was found in the linker histone variants H1.2, H1.3, and H1.4. So, for all tested linker histone variants, it was established that only serines were phosphorylated during interphase, but in mitosis, threonine residues were additionally phosphorylated. It was found that, during interphase, the human lymphoblastic T-cells had a proportion of H1.5 molecules monophosphoryated at a particular residue and a smaller proportion that was monophosphorylated on another residue. It was also found that the ratio of these two subgroups of H1.5 occurred in other cell types. During mitosis, it was found that H1.5 existed as two species with five phosphorylations either on Thr11, Ser18, Thr138, Ser173, and Ser189, or on Thr11, Ser18, Thr155, Ser173, and Ser189. Therefore, it was concluded that Thr138 and Thr155 of H1.5 can never be phosphorylated at the same time. There is support for this hypothesis [42], where the only phosphorylations to be found on H1.5 were at Thr138 and Thr155. If these two modifications occurred at the same time, it would be reasonable to expect that they would be found in equal abundance. However, whereas the H1.5 peptide with a phosphorylation on Thr155 was readily detected, that with a phosphorylation at Thr138 could only be detected after methanolic HCl was used to convert carboxylic groups to their corresponding methyl esters. Thus, the suggestion is that H1.5 exists as two separate species, with either Thr138 or Thr155 phosphorylated, but not both. Wisniewski et al. [43] (Table 1) identified phosphorylation at Thr138 on H1.5, but not at Thr155, even though, like Sarg et al. [41] and Garcia et al. [42], they used HeLa cells. Wisniewski et al. [43] agree with Sarg et al. [41] that Ser18 of H1.5 is phosphorylated, but could find no such modifications at Thr11, Ser173 or Ser189 in human or mouse tissue.

Table 1.   Alignment of chicken, mouse and human PTMs. Each PTM-containing sequence in humans has been aligned with the similar sequence in the other two species, which may or may not contain a PTM. Symbols: a, acetylation; d, deamidation; f, formylation; m, methylation; p, phosphorylation; u, ubiquitination; 2m, dimethylation; 2m/f, dimethylation and/or formylation; a/m, acetylation and/or monomethylation; a/f, acetylation and/or formylation; 2m/f, dimethylation and/or formylation. The ‘a-’ in the second column (first PTM location) refers to N-terminal acetylation. The data for human and mouse were taken from [43], and the data for chicken were taken from [44].
Chicken
 H101a-STAAPPAmKAKAKATKKK2m/fKKdNK
 H110a-STAAPAAKAKAKATKKK2m/fKKdNK
 H102a-STAAPSAKAKPKATKKK2m/fKKdNK
 H103a-ApTAAPAAKAKAKATKKK2m/fK2mKdNK
 H11La-STAPAAAKAKAKATKKK2m/fKKdNK
 H11Ra-ApTAAAAaKAKAKATKKK2m/fKKdNK
 H5a-pTpSApSPAAamKRpSpSTAKQ2m/fKNDR
Mouse
 H1.0a-TNpSAaKKKSDSAKQKEDK
 H1.1a-pSpTAASaKPaKmKAKKApSQKufKKa/mKNaK
 H1.2a-pSAAAAaKAKKmKRa/mKpSpSKaKufKKa/mKNafK
 H1.3a-STAAP2mKpTKKTRa/mKpSpSua/mKaKufKKa/mKNafK
 H1.4a-pSpTAAPKpTKKKRmKpSpSa/mKaKufKKa/mKNafK
 H1.5a-pSpTAEPKpSKKKmRmKpTSKKKKmKNK
Human
 H1.0a-TNSAKKSDSAKQKEDK
 H1.1a-pSTVASKPaKKPKKASQKfKaKKNK
 H1.2a-pSTAAAaKAKKARaKpSSauKaKfKaKKNaK
 H1.3a-STAAPaKpTKKKRaKpSSauKaKfKaKKNaK
 H1.4a-STAAPaKpTKKARaKpSSauKaKfKaKKNaK
 H1.5a-pSTAEPaKpSKKKRKTSaKKKKKNafK
Chicken
 H101aKKRRTAaKKpSKKTKKaKSAKSK
 H110aKKRRPAaKKpSKKTKKaKSTKSK
 H102aKKKKPAaKKpSKKTKKaKSTKSK
 H103aKKRKPAaKKpSKKTKKaKSTmKSK
 H11LaKKRKSAaKKpSKKTaKKKSTa/mKSK
 H11RKKRKSAaKKpSKKTaKKKSa/mKSK
 H5AKRGSTKAKKTKAKSaKKTA
Mouse
 H1.0TKRKAVAAKAKKKATKVK
 H1.1uKKKuKaKAATKKKKKSKKSK
 H1.2aKKmfKQGATpTTKaKAaKKKSaKS
 H1.3aKKmfKKGATTTKKAaKKKSKKSK
 H1.4aKKmfKKGAATTaKaKAaKKKSKKpSK
 H1.5aKKKKGTATKaKAaKKKSKKSK
Human
 H1.0TKRFKAATAKKKATKAKK
 H1.1a/fKaKKKSTTKKRKKNKKSK
 H1.2a/fKaKKKGAApTTKKpTaKKKpSKKSaK
 H1.3a/fKaKKKGAApTTKKAaKa/mKKSKKSK
 H1.4a/fKaKKKGATpTTKKAaKa/mKKpSKKpSK
 H1.5KKKKGpTKGTKKaKa/mKKSKKSK

Correlation of N-terminal tail PTMs with function

Sarg et al. [41] could not detect any N-terminal tail phosphorylations for H1.2, H1.3 and H1.4 for cells that were in the interphase part of the cell cycle. The reason given is that there are no (S/T)PXZ motifs in the N-terminal tails of those variants. This is in conflict with the work of Garcia et al. [42], who found phosphorylations on H1.2 at Thr31 and Ser36, on H1.3 at Ser37, and on H1.4 at Ser2, Thr4, Thr18, Ser27, and Ser36. In comparing the work of Wisniewski et al. [43] (Table 1) and Garcia et al. [42], it can be seen that: H1.4 is phosphorylated at Thr18, but not on Ser2 or Thr4; H1.3 is similarly phosphorylated at Ser37; H1.2 is not phosphorylated at Thr31. Thus, it is possible to conclude that phosphorylation of the N-terminal tails of the H1 variants does occur, but why do some researchers detect them but others do not? Before addressing this question, there is also the issue as to what variety of PTMs occur in the N-terminal tails. Sarg et al. [41] found that only H1.5 was modified in the N-terminal tail in human cells; more recent work by Wisniewski et al. [43] and Snijders et al. [44] (Table 1) has, excluding the N-terminus acetylations, identified eight N-terminal tail PTMs in cultured human cells and seven in Gallus gallus (hereafter ‘chicken’) erythrocytes, respectively. Although the overall number of these modifications is low, the density of modifications is much the same as in the rest of the linker histones. It is the shortness of the N-terminal tails that accounts for the low numbers. Therefore, it is now possible to say that the N-terminal tails do, in fact, contain a range of different types of PTM.

Returning to the issue of abundancy, Garcia et al. [42] had to use two techniques to increase the number of peptides with certain PTMs. First, protein digests were treated with propionylation reagent to convert monomethylated and endogenously unmodified amino groups on the side chains of lysine residues and N-termini to propionyl amides. Second, it was found that certain phosphorylated peptides (predominantly originating from the N-terminal tail) were of such low abundance that, in order to obtain stronger spectra, they were subjected to enrichment by immobilized metal affinity chromatography [45]. Prior to using this technique, Garcia et al. [42] converted the carboxylic groups to methyl esters with the use of methanolic HCl. This modification decreases the strength of binding of nonphosphorylated linker histones to the immobilized metal affinity chromatography column. They can then be washed off before eluting the phospho-linker histones. It should be noted that the practice of methyl esterification is currently not widely used, owing to problems with side reactions [46]. Deterding et al. [47], who only analysed the H1.4 linker histone variant in human and mouse tissue, found that, in both species, only Thr18 in the N-terminal tail region was phosphorylated. However, the signal for this in the human tissue was so noisy in comparison with signals for phosphorylated residues in the C-terminal tail that confirmation could only be obtained by reference to the mouse signal, which was less noisy. With high-mass accuracy mass spectrometers now readily available, this should be less of a problem in the future. We can perhaps, then, hypothesize that although N-terminal tail modifications of linker histones do occur, they may be less abundant than those of the globular domain and C-terminal tail. The degree to which they are less abundant remains to be established, as does the biological significance of that fact.

Correlation of C-terminal tail PTMs with function

Sarg et al. [41] found that H1.2, H1.3 and H1.4 had fewer phosphorylations in the C-terminal tail region than H1.5. In H1.2, Ser173 is phosphorylated, as are Ser189 in H1.3 and Ser172 and Ser187 in H1.4. Garcia et al. [42] found that H1.2 was phosphorylated at Ser173, Thr146, and Thr154. It is worth noting that these two latter modifications occur in TPXZ motifs and, if Sarg et al. [41] is correct, may reflect the fact that the cells are in mitosis. The H1.3 variant in Garcia et al. [42] was found to be phosphorylated at Ser189, Thr147, Thr155, and Thr180. The latter threonine is in a non-CDK consensus site, and seems to be an anomaly. The occurrence of phosphorylated Thr147 and Thr155 in the C-terminal tail of H1.3 probably has the same explanation as their occurrence in H1.2. Deterding et al. [47] identified the same H1.4 C-terminal tail phosphorylations in human and mouse tissue as Sarg et al. [41]. As can be seen from Table 2, Wisniewski et al. [43] detected no phosphorylation on Ser172 of H1.4. For H1.2 and H1.3, the work of Wisniewski et al. [43] agrees with that of Garcia et al. [42], noting that in Table I of the former paper the phosphorylation on Thr173 of H1.2 is a typographical error (should be Ser174).

Table 2.   Phosphorylations of the human linker histones. Those modifications that have been identified more than once in the papers referred to are shown. The numbers in parentheses refer to the appropriate references.
H1.2 (P16403)H1.3 (P16402)H1.4 (P10412)H1.5 (P16401)
pS36 [42,43], pT146 [42,43]pS37 [42,43], pT147 [42,43]pT18 [42,43,47], pS172 [41,42,47]pS18 [41,43], pT138 [41–43]
pS173 [41–43]pS189 [41–43]pS187 [41–43,47]pT155 [41,42]

Table 2 lists the phosphorylations that have been detected more than once in the research described above. It therefore represents those sites that are most likely to be modified at reasonable levels of abundance.

Most mass spectrometry (MS) analysis of PTMs has been performed on cultured cell lines. It has been shown that methylation can be readily detected in tissue, but is extremely rare in cultured cells [43]. Other potential problems with cultured cells are discussed later.

Analysis of other PTMs

It is now accepted that acetylation and methylation of the core histones are key regulators of transcription. Although phosphorylation of the linker histones has attracted the most attention, recent results from various MS analyses have shown that acetylation and methylation are also key modifications of the linker histones.

Lysine and N-terminus acetylations

Acetylation of the Nα-terminus involves the cotranslational cleavage of a methionine, followed by acetylation of the second residue of the N-terminal tail. Initial, non-MS work showed that only H1.0 in human cells and H5 in avian cells existed in forms that had the Nα-terminus both acetylated and unacetylated [48,49]. However, it has now been shown that the former modification not only occurs on all the linker histone variants in human and chicken [42–44], but that is also the most abundant type of acetylation. Linker histones do exist with their Nα-termini unacetylated, but are significantly fewer in number. Although the function of Nα-terminus acetylation is unclear, it was noted in earlier work [48,49] that whereas the ratio of acetylated to unacetylated Nα-termini remained constant in avian H5 erythrocytes from newly hatched and adult chickens, it increased for H1.0 in ageing rat tissues. As H1.0 is associated with differentiation and is most abundant in terminally differentiated cells, there may well be a correlation between Nα-terminus acetylation and differentiation. Not all methionines are cotranslationally cleaved and can therefore become acetylated. This process is not widespread, but it has been shown to be present in recent work [43,44] (Table 1).

Garcia et al. [42] found that H1.2, H1.3 and H1.4 in human cells all had just one site of lysine acetylation, and on the same residue, Lys64 (or Lys65, depending on the variant). Considerably more acetylations – up to nine – were found in H1.4 [43] (Table 1). The globular domain was found to contain the largest number of acetylations: Lys52, Lys64, Lys85 and Lys97; all of these are thought to be involved with DNA binding [43]. The abundance of lysine methylation can be attributed to the fact that certain types of human cell were rapidly proliferating. In mouse tissue, the spleen was found to contain the most acetylations, because lymphopoiesis is associated with rapid cell division. In mouse tissue containing mostly differentiated cells, e.g. liver, the number of acetylations was much lower.

Lysine methylation

Methylation of lysine in linker histone proteins has been reported in human HeLa cells [42,43], although there is a difference in the number of identified sites. Garcia et al. [42] found that, in H1.4, Lys26 and Ser27 were simultaneously methylated and phosphorylated, respectively. The point is made that the aforementioned residues occur in the sequence KARKSAGA (residues 23–30), which is similar to one found in the core histone H3 (VARKSAPA, residues 25–31). Within H3 there are well-known adjacent methylation and phosphorylation sites at Lys9-Ser10 and Lys27-Ser28 that are involved with transcription. Thus, the same argument is made for H1.4, by virtue of it, too, having adjacent methylation and phosphorylation sites. There is support for these assertions [49]; however, Wisniewski et al. [43] (Table 1) found no methylations in this region of H1.4, or of the other variants. Putative sites of methylation were identified at Lys169 in H1.4 and H1.5, or Lys170 in H1.3. In mouse, the H1.4 variant has no modifications at Lys26, and at position 27 there is an alanine, rather than a serine [43].

Ubiquitination and formylation

For the first time, ubiquitinations of the histone linker protein were identified by MS [43] (Table 1). It was found that Lys46 was ubiquitinated in H1.2, H1.3 and H1.4 in human HeLa cells, but not in MCF7 cells. In mouse tissue, Lys116 of H1.1 and Lys46 of H1.2 and H1.3 in the spleen were the only sites of ubiquitination. The fact that both cell lines were cultured in the same growth medium, and have the same doubling time, increases the probability that these ubiquitinations are unique for HeLa cells. Another putative novel modification found is that of formylation [43]; H1.1, H1.2, H1.3, H1.4 and H1.5 in human MCF7 cells were all found to be formylated. Whereas H1.5 was uniquely formylated on Lys88, the others were similarly modified on Lys90 (H1.2 number). In mouse tissue, the most frequently occurring formylation site was Lys63. Formylation of lysines has been shown to arise as a result of oxidative damage to DNA [50]. Snijders et al. [44] (Table 1) identified a single site of lysine dimethylation at Lys71, but were unable to distinguish between dimethylation and formylation.

Perturbation of phosphorylations by external mechanisms

It has been clearly shown in several studies that phosphorylation can be imposed by external influences [41,47,51]. This is an important phenomenon, and means that those processes will be able to influence the cell cycle.

Garcia et al. [42] found that growing T. thermophila cells had site-specific higher levels of phosphorylation than when they were being starved. Phosphorylated Thr47 was enriched in growing cells by a factor of seven as compared with starved cells. Similarly, phosphorylated Thr35 was also found to be enriched by a factor of four in growing cells. It is perhaps important that these two residues occur in (S/T)PXZ motifs (as defined). It was found that in Drosophila melanogaster embryos, phosphorylated Ser11 was associated with mitosis and that the proportion of this post-translational modification decreased as those embryos aged [52]. These experiments clearly show that the amount of phosphorylated H1 is a function of cell activity. Villar-Garea and Imhof [52] concluded that, in mammalian cells, phosphorylation in mitosis only occurs in the N-terminal tail. However, it has been shown that, during mitosis, phosphorylations also occur in the C-terminal tail, and in (S/T)PXZ motifs [41].

Deterding et al. [47] analysed human UL3 cells (derived from the osteosarcoma cell line U2OS) treated with dexamethasone, CVT313, or CGP74514 (dexamethasone is a synthetic glucocorticoid used in the treatment of autoimmune diseases; CGP74514 and CVT313 are CDK1 and CDK2 inhibitors, respectively), and looked at the phosphorylation state of the linker histones using MS. It was found that treatment with all of these compounds reduced the global level of phosphorylation of the H1.2 and H1.4 isoforms. Although the work did not go so far as to establish site-specific patterns of phosphorylation related to compound and dosage, it did establish, by the use of antibodies, that the level of phosphorylated Thr18 in the N-terminal tail of H1.4 was reduced by treatment with any one of these three compounds.

In the examples discussed, extensive use was made of cultured cells. There is, of course, nothing wrong with this in the substantial majority of cases. However, culturing cells may have an impact on overall PTM patterns. In particular, differences may arise in the structural and biochemical properties of a cultured cell (and hence PTM patterns), particularly when the cells are grown on a monolayer 2D medium. Normal cells grown in such a medium can display a nuclear structure that is different to their in vivo structure [53,54]. Use of a 3D culture medium better mimics the extracellular matrix [53], and the cells should therefore have a nuclear structure that is more representative of the in vivo structure. If the nuclear structure of cultured cells can be altered, then there will be a concomitant change in the biochemistry of those cells [54].

Existence of global PTM patterns in different cell types

The strong evidence emerging is that specific PTM patterns occurring on DNA and particular sets of proteins can be correlated with cell type. The inference from this is that there will be a change in a cell’s PTM pattern when it progresses from a normal to diseased state, and that, accordingly, such changes can be detected and made the target of clinical intervention [55,56]. However, although changes in the PTM patterns of particular proteins between normal and diseased cells have been detected [55,56], can the concept be taken to the lower level of chromatin? This has already been shown to be the case in three sets of mouse cells [57]. A proportion of murine embryonic stem cells, embryonic fibroblasts and embryonic carcinoma cells were grown in standard cell growth medium, with the remainder having trichostatin A, an HDAC inhibitor, added, the aim being to mimic disease-induced hyperacetylation of histones. Two changes were detected: (a) PTM patterns alter for ‘diseased’ cell lines; and (b) those cell lines have unique and specific PTM patterns. The PTMs that were being monitored resided on the H3 and H4 core histones. However, there is nothing to suggest that the linker histones should not display the same global property, and such changes involving just phosphorylation were discussed in an earlier section. The process of disease-induced alteration of global PTM patterns has also been observed in human colon adenocarcinoma cell lines [58]. MS is of fundamental importance when it comes to detecting combinations of PTMs on a single protein. This has been demonstrated on human embryonic stem cells [59].

Table 1 shows the PTMs detected in human, mouse and chick cells [43,44]. From the data for chick cells, it can clearly be observed that six of the linker histones have identical PTMs at amino acids 71, 84, 147, and 189. Unlike human linker histones, the chick variants have very similar sequences, and it would be easy to dismiss this observation with the argument that near-identical sequences will inevitably have the same PTM pattern. However, this line of argument would ignore two important facts. First, the chick linker histones can – like their human counterparts – be associated with specific and different functions; and, second, the cells, being erythrocytes, are terminally differentiated. It is therefore possible to say that the PTM patterns in Table 1 for chick cells can be identified as being unique for terminally differentiated chick erythrocyte cells. However, as previously mentioned, although MS can detect many modifications, it does have restrictions, such as difficulty in distinguishing PTMs that have near-identical masses [44].

It was mentioned earlier that mouse tissue with the higher replication rate has higher levels of linker histone acetylation. This can be taken as evidence that unique linker histone PTM patterns also exist in live tissue, and not just in cultured cells [43]. It is possible to come up with a list of PTMs that are either absent in MCF7 cells and present in HeLa cells, or present in MCF7 cells but missing in HeLa cells (Table 3). As mentioned earlier, the two cell lines were grown in the same media, so it is clearly possible to distinguish the two human cell lines by comparison of the PTMs on their linker histones.

Table 3.   PTMs that uniquely identify human MCF7 from HeLa cells. ‘+’ indicates a modification that is present on MCF7 linker histones but is missing from HeLa linker histones. ‘−’ indicates a modification that is missing from MCF7 linker histones but is present on HeLa linker histones. The symbols in italics are as defined for Table 1.
MCF7 PTMs compared with HeLa PTMs
H1.5: +aK17
H1.1: −aK22
H1.2, H1.3, H1.4: +aK34
H1.2, H1.3, H1.4: −uK46
H1.2, H1.3, H1.4: −aK52
H1.1, H1.2, H1.3, H1.4: −aK64
H1.1, H1.2, H1.3, H1.4: −aK85
H1.5: +a/fK88
H1.1, H1.2, H1.3, H1.4: −fK90
H1.1, H1.2, H1.3, H1.4: +aK97
H1.2: −pT146
H1.2: +pT165
H1.2: +aK169
H1.4: +aK169

Conclusions

The evidence accumulating from MS and other biophysical experiments considerably strengthens the hypothesis that not only can the linker histone variants be associated with specific functions, but PTMs thereon can also uniquely identify particular cell types. Indeed, this is now becoming the accepted paradigm [60]. Those functional capabilities even, as in the case of H1.2, have an extranuclear reach. PTMs modulate the range of functions covered by the linker histone variants and, by analogy with the core histones, each of those functions will have a distinct PTM signature.

The extraction of the H1 variants from cells or tissue has the potential to alter PTM states. It is therefore necessary that gentle procedures should be used. Acid extraction of H1, although efficient, can instigate the reversal of labile PTMs, such as histidine phosphorylation [61–63]. A range of different acids have been used to extract linker histones, including perchloric acid [43,52], sulfuric acid [42,64], and hydrochloric acid [65]. Extraction by salt is gentler and just as efficient at isolating H1 histones [44,63]. Although it is the extraction of the linker histones from tissue and cell cultures that has a high potential to alter PTMs, purification of the extracts can be considered to be a benign step in the process of isolation. Purification of histones in general can involve a myriad of processes, and these have been discussed in detail elsewhere [66].

It was found in one case that phosphorylation of threonines in H1.5, namely Thr138 and Thr155, was associated with cells in mitosis [41]. Some support was provided for this principle [42], where both of these residues were found to be phosphorylated, although Thr138 only with some difficulty. In another case, it was found that only Thr138 was phosphorylated [43]. There is a clear conflict here, so is it possible to distinguish the respective cases? In the first case, the cell lines were specifically treated to put them into mitosis; this was not so in the second and third cases. However, in two cases [42,43], it seems unlikely that there would have been a significant number of cells in mitosis. In addition, in MS experiments, absence of a condition is not proof of its nonexistence.

Whereas, initially, it was found that PTMs in the N-terminal tail of most of the H1 variants did not occur – an exception being H1.5 [41] – it is now clear that there are, in fact, numerous modifications, although they seem to be less abundant [43,44]. Acetylation of the N-terminus of H1 is the most abundant modification, although it has been shown that the unacetylated form does exist [43,44]. There seems to be no consensus on the significance of N-terminus acetylation. However, as the amount of H1.0 with an acetylated N-terminus has been observed to increase in ageing rat tissue [48,67], and given that H1.0 is most abundant in terminally differentiated cells, there may be a link between N-terminus acetylation and differentiation.

From work on human cells and mouse tissue [43], it can be clearly seen that the amount of acetylated H1 is a function of the replication rate, with most acetylations occurring in rapidly replicating tissue, and the least in the most slowly replicating tissue. Confirmation of this comes from work on chicken erythrocytes [44], where it was found that there are relatively few acetylations in the chicken H1 variants. This is because the erythrocyte sample material comprises cells that are largely terminally differentiated.

Phosphorylation is correlated with growth rates [64] and can be significantly increased. The addition of compounds that influence the cell cycle will cause changes in the levels of phosphorylation of the linker histone isoforms [47].

Particular cell lines can be identified with particular patterns of PTMs on the core and linker histones [43], and variations in those patterns – having been associated with oncogenic progression [68] – are primary candidates for pharmacological intervention.

Taken as a whole, the data from the experiments discussed herein clearly show that it is possible to associate specific PTM patterns in the linker histones with particular functions, and that unique patterns of PTMs exist for diseased cells when compared with normal cells, and between cells of different types.

Acknowledgements

We would like to thank A. Evans for his advice on cell growth rates and S. Lambert for his advice and assistance in the preparation of the chicken linker histones. Both of the aforementioned are based in the School of Pharmacy and Biomolecular Sciences, Liverpool John Moores University.

Ancillary