1 The Simple Question

Let’s be more precise in the question we want to ask, and narrow the general “why do living creatures need metal ions?” to “why does the stomach ulcer-causing bacteria need nickel ions?”. This relatively simple question is a consequence of scientific curiosity, inspired by the discovery of Helicobacter pylori (H. pylori). Now, thirty years after Barry Marshall drank a suspension of the bacteria to prove that it is the cause of stomach ulcers, it is obvious that is the Gram-negative H. pylori, which colonizes the gastric mucosa in humans, causing acute and chronic gastritis, peptic ulcer disease, gastric carcinoma, and gastric lymphoma (Kusters et al. 2006). It is estimated that about 40–50% of the adult population in developed countries, and up to 90% of adults in developing countries can be infected with H. pylori and new, non-antibiotic treatments are still being looked for.

Coming back to the topic: why does H. pylori need Ni²⁺ ions? What happens when it doesn’t get any, and how can we make it happen?

2 The Complicated Analysis

The answer starts in a relatively simple way – H. pylori produces an assortment of factors in order to adapt to the extremely acidic environment of the stomach. We have to keep in mind, that although the bacterium successfully inhabits the extremely low pH of the stomach, it multiplies in an environmental pH from 6.0 to 8.0 and cannot survive at pH below 4.0 (Morgan et al., 1987).

To survive in those unwelcoming living conditions, H. pylori depends on two nickel-containing enzymes, urease and hydrogenase. The dinuclear Ni²⁺-containing urease accounts for about 10% of the soluble cellular proteins (Bauerfeind et al., 1997), catalyzes the hydrolysis of urea into carbon dioxide and ammonia and therefore neutralizes the low gastric pH around the bacteria (Scott et al., 2002). The activity of urease critically depends on the availability of nickel – a functional urease complex requires 24 Ni²⁺ ions (Hawtin et al., 1991). Another factor is a membrane bound [NiFe] hydrogenase, which permits respiratory based energy production for the bacteria in the mucosa (Olson & Maier, 2002). Most of the bacterium’s metal metabolism is centered upon the expression and maturation of those two Ni²⁺- dependent enzymes (which is not a surprise, because the task is quite challenging – the concentration of Ni²⁺ in human plasma is as low as 0.44 nM) and a set of at least 22 accessory proteins involved in the homeostasis of nickel is needed for their activation (Mehta et al., 2003). Understanding this mechanism is a crucial basis for developing new, highly specific treatments that would aim at the “weak point” of this machinery. Since the number of people resistant to standard therapy (two antibiotics and a proton pump inhibitor) is rapidly increasing (over 20% do not respond to the therapy at all), new solutions in the treatment of ulcers are still being looked for. One of such therapies is a group of drugs based on bismuth (III) salts. Currently, there are three different salts available on the market – bismuth subsalicylate (in Pepto-Bismol), colloidal bismuth subcitrate (in De-Nol) and ranitidine bismuth citrate are commonly used for the treatment of peptic ulcers, caused by the presence of H. pylori. How do they work? It is tempting to think that Bi³⁺ is able to displace Ni²⁺ from its binding sites. Which proteins look as if they could bind both metals? How can we find out how bismuth ions work, without getting lost in the complicated network of the large and numerous nickel-binding proteins (Figure 15.1)?

What should we start with, trying to answer our ‘simple’ (now it sounds ironic…) question? A good starting point would be a protein that has potential good metal binding sites in unstructured regions.

HspA, Hpn and Hpn-like are good examples of such.

View Full Size

The complex system of nickel homeostasis in H. pylori

• Download Figure
• Download figure as PowerPoint slide

Heat Shock Protein A (HspA) is a GroES homolog involved in the folding of polypeptide chains into oligomeric complexes. In the case of H. pylori, HspA has a second function, being involved in the homeostasis of nickel. It delivers nickel to appropriate proteins in a Ni²⁺-deficient environment and takes part in the detoxification through the sequestration of excess nickel. Under normal conditions, HspA delivers Ni²⁺ to proteins that have a higher affinity towards this metal than HspA itself. In H. pylori, the protein consists of 118 amino acids divided into two domains: an N-terminal domain (domain A, residues 1–90), which is homologous with other GroES bacterial proteins, and an unstructured histidine and cysteine-rich C-terminal domain (domain B, residues 91–118), which other GroES-like proteins do not have (Suerbaum et al., 1994) (Figure 15.2-A). This is a good indication that this is the domain we want to take a closer look at in our ‘simplified’ approach.

Two other proteins that could be ‘suspected’ of being excellent nickel binders are the extremely His-rich Hpn and Hpn-like, two of Helicobacter pylori’s cytoplasmic proteins involved in the homeostasis of nickel. Almost half of Hpn’s amino acid sequence (28 out of 60 residues) are histidines (Figure 15.3). Hpn-like, highly similar in its primary sequence to Hpn, contains some additional polyglutamine repeats (25% of its sequence are histidines, while 42% of the amino acid residues are glutamines). Both proteins are not only urease and hydrogenase nickel accessory proteins but also important factors in nickel (and other metals) storage and detoxification (Li & Zamble, 2009). Hpn is abundant in the cytoplasm of H. pylori and accounts for approximately 2% of all proteins synthesized by the bacterium (reviewed in: Mulrooney & Hausinger, 2003; Kaluarachchi et al., 2010).

Now, having those proteins in mind, let us again ask: what happens, when something ‘goes wrong’ with nickel distribution in H. pylori? To answer the simple question, it will be necessary to make several simplifications of the complex system.

3 The Simplification

In this case, the simplification is based on working with model systems – peptides, which are representative unstructured fragments of the studied nickel binding proteins, ‘suspected’ of being the metal binding site. Simple: the idea is to work with the metal binding site, not with the full protein, and to precisely determine the affinity binding constants to each of the potential metal binding sites or to investigate peptide recognition. Or, even simpler: we chop up the protein into smaller pieces and we check where the metal binds most strongly or we check if the two regions interact with each other.

This simplification has several limitations that we have to be aware of. It would not be particularly useful if the studied protein region had a pre-defined structure, or was a membrane protein (I will come back to this point later, now the plan is to make things simple; the verification process will follow).

Let us first focus on the C-terminal part of HspA. Its eight His and four Cys residues among 27 amino acids (GSCCHTGNHDHKHAKEHEACCHDHKKH) make it a tempting nickel binder. Keeping in mind, that the N-terminal domain has the function of a heat shock protein, and is not involved in nickel homeostasis, and that the C-terminal domain (absent in other species) is unstructured and very probable to chelate metal ions, it is reasonable to focus on domain B only. The shorter the studied region, the more precise thermodynamic data we are able to obtain about its metal complex (not even to mention the lower costs and easier handling of the sample …).

Indeed, Ni²⁺ ions make a thermodynamically strong square planar complex with two sulfurs of neighboring cysteine residues and to the amide between them. Looking at the interaction of bismuth with the same model system, at least one of the ways how bismuth might inhibit nickel-dependent proteins of this bacterium becomes obvious – thermodynamically, bismuth complexes are over ten orders of magnitude more stable than nickel ones (Rowińska-Żyrek et al., 2009). Moreover, Bi³⁺ is able to displace Ni²⁺ from its Cys-rich binding sites (Rowińska-Żyrek et al., 2010), disrupting the homeostasis of nickel and/or causing nickel deficiency during post-translational modifications of urease and hydrogenase.

View Full Size

A. A scheme of Heat Shock Protein A from H. pylori

B. A suggested scheme of the structural rearrangement of HspA upon Bi³⁺ addition; only the front subunit is highlighted (Cun et al., 2008)

• Download Figure
• Download figure as PowerPoint slide

The finding was confirmed by the biological study, which showed that Bi³⁺ induces the changes in quaternary structure of the protein, making it refold from a native heptamer to a dimer (Cun et al., 2008). The binding is irreversible at physiological pH, which clearly indicates that bismuth may inhibit the biological functions of HspA (Cun et al., 2008).

How about the abnormally histidine-rich protein, Hpn? It mostly exists as an unstructured multimer in solution, with each 7 kDa monomer binding 5 nickel ions at pH 7.4 (Ge et al., 2006). Bacteria lacking Hpn, cultured in vitro have an increased response to therapeutic forms of bismuth (Mobley et al., 1999). How does this happen, what are the details of Ni²⁺ and Bi³⁺ coordination to this protein?

A striking feature of Hpn, apart from the impressive amount of histidines, is a variety of available binding sites for metal ions. Nickel, native to the protein, is able to bind to the ACTUN motif in an albumin-like binding (MAH), to the His-rich motif (HHH) and to the cysteine thiols (CC). Since Hpn does not have a defined structure and all of the mentioned sites are available for binding, this particular protein is an excellent example based on which we can debate on the thermodynamic preferences of nickel binding. What is the outcome of the debate? The N-terminal part of Hpn binds Ni²⁺ ions in the same way as native human albumin does, forming nickel and copper complexes involving a {NH₂ 2N^–, N_im} donor set. Two other fragments with a -CC- motif (Ac-CCSTSDSHHQ-NH₂ and Ac-EEGCCHGHHE-NH₂) derived from the C-terminal part of Hpn, bind nickel, forming very stable planar complexes with both peptides, due to the {2S^–,N^–} binding mode (Rowińska-Żyrek et al., 2013).

The ‘simplified’ approach allowed an understanding of the behaviour of a polyhistidyl fragment of Hpn (Ac-THHHHYHGG-NH₂) in a complex with Ni²⁺. The coordination properties of this so-called wild-type fragment were compared with those of its six analogues, in which consecutive residues (His or Tyr) were replaced by Ala (Ala-substitution or Ala-scan approach) resulting in Ac-TAHHHYHGG-NH₂, Ac-THAHHYHGG-NH₂, Ac-THHAHYHGG-NH₂, Ac-THHHAYHGG-NH₂, Ac-THHHHAHGG-NH₂ and Ac-THHHHYAGG-NH₂ peptides, respectively. This approach allowed to conclude that the fourth His residue is critical for the binding of Ni²⁺ and that the effectiveness of binding varies even if the substituted amino acid does not directly bind. Moreover, we showed that the metals cannot bind to four consecutive histidines (Witkowska et al., 2012).

View Full Size

Binding modes of nickel to the Hpn sequences discussed above – HHYH, MAH and CC

• Download Figure
• Download figure as PowerPoint slide

One of the most useful features of the “peptide-based approach” is the precision with which we can compare the thermodynamic differences of the different binding sites. Among the three different parts of Hpn mentioned above, it is the C-terminal part of the protein with the “CC” sequence which binds nickel most effectively. The His-rich fragment binds nickel more efficiently at lower pH than the N-terminal part does; at physiological pH (7.4), we observe the contrary.

Another fun fact about Hpn and Hpn-like: the more glutamine residues they have, the stronger they coordinate metal ions. The ‘simplified’ approach was used to study the interactions of Ni²⁺ with several N-terminal domains of Hpn and Hpn-like proteins from H. pylori, each with a different number of glutamine residues, but with the same albumin-like metal binding mode. Experimental results, in very good agreement with theoretical findings, lead to the not-at-all obvious conclusion that the stability of metal complexes distinctly increases with the number of glutamine residues present in the peptide, although glutamine side chains do not directly take part in coordination. Most probably forms a network of hydrogen bonds which protects the metal binding site from water molecules. This peculiar finding allows one to look at polyglutamine sequences, not only the ones present in some bacterial chaperons but also those involved in several neurodegenerative diseases, from a new perspective (Chiera et al., 2013).

4 The Simple Answer

Helicobacter pylori depends on urease and hydrogenase to survive in the acidic environment of the human stomach. Both of the two enzymes need nickel ions, which is why the bacteria express numerous proteins that ensure the proper distribution and storage of Ni²⁺. Moving in the world of H. pylori’s protein interactions with metals, using model systems – unstructured protein fragments, which are ‘suspected’ of being metal-binding sites, we can point out the regions with the highest affinity towards nickel and prove that the component of commonly used antiulcer drug, bismuth, may inhibit nickel dependent proteins of this bacterium – thermodynamically, bismuth complexes are over ten orders of magnitude more stable than nickel ones and Bi³⁺ is able to displace Ni²⁺ from its Cys-rich binding sites (Rowińska-Żyrek et al., 2010), disrupting the homeostasis of nickel and causing nickel deficiency during post-translational modifications of urease and hydrogenase.

‘Making the complicated simple, awesomely simple’– a series of cases

Clearly, H. pylori related studies are not the only ones to which the “simple question-complicated analysis-simplification-the simple answer” deduction scheme can be applied. Even more than that – the system doesn’t even have to be a protein. Examples? With pleasure – a ribozyme example. The “divide and conquer” policy told us a lot about the correlation between the structure and function of one of the only three known human small catalytic RNA s, the CPEB3 ribozyme (Figure 15.4A).

Until recently, ribozymes (‘enzymes’ composed solely of RNA) were found only in lower organisms. The discovery of the human CPEB3 ribozyme, related to the Hepatitis Delta Virus (HDV, Figure 15.4B) one, was an enormous breakthrough.

This ribozyme is a highly conserved, mammalian, self-cleaving, non-coding RNA located in the second intron of the cpeb3 gene (Salehi-Ashtiani et al., 2006). This gene encodes a cytoplasmic polyadenylation element binding protein that promotes the elongation of the polyadenine tail of messenger RNA and mediates germ cell development, synaptic plasticity, learning and memory; it has also been suggested to adapt prion-like conformations. The CPEB3 protein is rather well studied, yet surprisingly little is known about the CPEB3 ribozyme itself. Most of the available information is based on comparative studies with the structurally and biologically related HDV ribozyme. Despite having divergence in base sequences, both the viral HDV and the human CPEB3 were predicted to form the same base pairing interactions, which result in a pseudoknot structure (Webb et al., 2009). How to verify the formation of the predicted nested double pseudoknot in solution and how to get an understanding of the structure-function and metal ion binding-function relationships of this 67 nucleobase-long ribozyme? The NMR structure of the whole construct is highly desirable (and highly complicated to obtain), as well as NMR-monitored Mg²⁺ titrations, to characterize specific metal binding sites and structural changes within the ribozyme upon addition of the metal ions.

NMR proton resonance assignment of the CPEB3 ribozyme was a huge challenge because of the severe spectral overlap in the [¹H,¹H]-NOESY spectra, which is mainly due to the large size of the ribozyme. All assignments were made in spectra of partially deuterated CPEB3 RNA, in which line widths were drastically improved compared to the spectra of natural abundance samples. The assignment was supported by using different labelling schemes for the CPEB3 RNA, which helped to select for [¹H,¹H]-NOESY or [¹H,¹⁵N]-HSQC correlations belonging to certain nucleotides. Four samples with only one NMR-visible (natural abundance) nucleotide (with the remaining three being fully deuterated ones) were also analyzed (Rowińska-Żyrek et al., 2014a).

Further NMR analysis showed that the global fold of the CPEB3 ribozyme, encompassing 80–90% of the predicted base pairs, is formed in monovalent ions alone. Low millimolar concentrations of Mg²⁺ promote a more compact fold and lead to the formation of additional structure in the core of the ribozymes, which contains the inner small pseudoknot and the active site. Several Mg²⁺ binding sites, which are important for the functional fold, appear to be located in corresponding locations in the HDV and CPEB3 ribozyme, demonstrating the particular relevance of Mg²⁺ for the nested double pseudoknot structure.

Still, no clue about structure? Keep it simple; apply the ‘divide and conquer’ strategy – use three small model constructs for the P1 and P2 helices and for the P4 hairpin (Figure 15.4C). This simple idea was an absolute hit – results obtained for the three separate regions allowed solving their structures and facilitated the assignment of the full-length CPEB3 ribozyme spectra (Rowińska-Żyrek et al., 2014b).

Another example of an approach that makes the complicated simple? Understanding the interactions of Zn²⁺ with zincophores and zinc transporters from Candida albicans (C. albicans), the most common cause of fungal infections in humans.

C. albicans, although usually a commensal fungus, it is the most common cause of candidiasis – a condition that encompasses infections that range from superficial, chronic to systemic and potentially life-threatening candidemia. The frequency of drug-resistant invasive mycoses has increased significantly over the past three decades, and is becoming a serious medical problem not only in immunosuppressed patients. As a result, new antifungal drugs and ways to specifically deliver them to the fungal cell are being looked for. One of the biggest obstacles in finding such comes from the fact that fungi share essential metabolic pathways with humans. One significant difference in the metabolism of those two cells that can be challenged when looking for possible selective therapeutics is the uptake of zinc (crucial for fungal survival and virulence).

View Full Size

Secondary structures of (A) the genomic HDV ribozyme (B) the CPEB3 ribozyme (C) the small model constructs P1, P2 and P4 used for resonance assignment; the nucleotides added to the natural sequences are shown in grey

• Download Figure
• Download figure as PowerPoint slide

Candida albicans relies on a mechanism of zinc uptake. It is based on a secreted protein, which specifically binds Zn²⁺, Pra1, the so-called ‘zincophore’, a small, 299 amino acid secreted zinc-binding protein, which can sequester this metal from the environment and re-associate with the fungus via a co-expressed, genetically-linked membrane transporter, Zrt1. Citiulo et al., (2012) elucidated the mechanism of C. albicans zinc acquisition from host cells: (1) after the host cell invasion Pra1 (pH-regulated antigen 1, the previously mentioned zincophore), is expressed due to the alkaline pH and low amount of soluble zinc of the intracellular environment (Sentandreu et al., 1998; Outten & O’Halloran, 2001); (2) the protein is secreted and released from the fungal cell surface, predominantly in the hyphal form; it is required for hyphal extension and causes endothelial damage of the host (De Bernardis et al., 1998); (3) it binds host cellular zinc (either free cytosolic or bound to host protein); and (4) returns to the fungal cell via physical interaction with Zrt1, a membrane transporter, to deliver the bound metal ion (Figure 15.5) (Citiulo et al., 2012).

How to understand the bioinorganic chemistry of this process, to point out the Zn²⁺ binding sites in both Pra1 and Zrt1 and understand the thermodynamics of the Pra1-Zn²⁺-Zrt1 interaction?

As you might already be guessing, our approach involves working on both full-length proteins and model systems (unstructured parts of proteins) in order to identify those regions in Pra1, to which bind zinc with the highest affinity and those which are recognised by Zrt1. Several unstructured regions of Pra1 and Zrt1 can be used to perform a structural and thermodynamical analysis of their complexes with Zn²⁺, being a stepping stone towards finding new, fungus-specific treatments based on parts of zincophores coupled with an imidazole- or triazole-based antifungal drugs.

Verification of the simplification procedure? The best one would be performing growth promotion studies, indicating whether these Pra1 fragments are able to transfer zinc into fungi, and therefore act like zincophores.

View Full Size

Schematic model of C. albicans zinc scavenging from host cells. After invasion of the host cell, Pra1 is expressed and secreted. It binds zinc, either in the form of free Zn²⁺ (extremely sparse in the cellular pool) or from zinc-binding proteins of the host. Reassociation with C. albicans cell surface and Zn²⁺ transport into the cell occurs via a Pra1-Zrt1 interaction

• Download Figure
• Download figure as PowerPoint slide

‘Peptide-based’, ‘divide and conquer’, however, we call the approach, if we are big fans of it or not, what is clear are two things: first, it was the key to several major discoveries – it would take time and severely abuse the dimensions of this book to describe all in brief (e.g., this simple approach allowed to elucidate the impact of metal ions on the molecular basis of Alzheimer’s, Parkinson’s or prion disease. Cu²⁺ and Fe³⁺ usually bind to specific sites in beta-amyloid, a-synuclein and prion proteins, resulting in misfolding and oxidative stress, which is the trigger for the diseases).

And second, it is of major importance to be aware of the obvious limitations of the method. A pre-defined structure of the biomolecule or a tertiary interaction with another group is a serious limitation.

Having started the chapter with a quotation, I will also finish it with one. I think one attributed to Albert Einstein fits: “Make everything as simple as possible, but not simpler” (Robinson, 2018).