The history and rationale of using carbonic anhydrase inhibitors in the treatment of peptic ulcers. In memoriam Ioan Puşcaş (1932-2015).

Abstract Carbonic anhydrase (CA, EC 4.2.1.1) inhibitors (CAIs) started to be used in the treatment of peptic ulcers in the 1970s, and for more than two decades, a group led by Ioan Puşcaş used them for this purpose, assuming that by inhibiting the gastric mucosa CA isoforms, hydrochloric acid secretion is decreased. Although acetazolamide and other sulfonamide CAIs are indeed effective in healing ulcers, the inhibition of CA isoforms in other organs than the stomach led to a number of serious side effects which made this treatment obsolete when the histamine H2 receptor antagonists and the proton pump inhibitors became available. Decades later, in 2002, it has been discovered that Helicobacter pylori, the bacterial pathogen responsible for gastric ulcers and cancers, encodes for two CAs, one belonging to the α-class and the other one to the β-class of these enzymes. These enzymes are crucial for the life cycle of the bacterium and its acclimation within the highly acidic environment of the stomach. Inhibition of the two bacterial CAs with sulfonamides such as acetazolamide, a low-nanomolar H. pylori CAI, is lethal for the pathogen, which explains why these compounds were clinically efficient as anti-ulcer drugs. Thus, the approach promoted by Ioan Puşcaş for treating this disease was a good one although the rationale behind it was wrong. In this review, we present a historical overview of the sulfonamide CAIs as anti-ulcer agents, in memoriam of the scientist who was in the first line of this research trend.


Introduction
Carbonic anhydrase (CA) is a ubiquitous metalloenzyme catalysing the reversible conversion of carbon dioxide to bicarbonate and has various biological functions in bacterial metabolism, respiration, photosynthesis, renal acidification, gastric acid and pancreatic bicarbonate secretion as well as bone metabolism, to mention only the most important ones. After a historical overview of CA research, this article will summarize the contribution of Romanian researcher Ioan Puşcaş to the multi-decade CA story, focusing on the relationship between gastric CAs and peptic ulcer disease, followed by a presentation of main achievements of bacterial CAs.

History of CA
Physiologists and chemists from the early decades of the 20th century were interested to find out how carbon dioxide (CO 2 ) is transported in blood 9,10 . The timeline of the experiments is presented in Table 1. Many important personalities of modern physiology were involved and the research culminated in 1933, when Norman Urquhart Meldrum and Francis John Worsley Roughton first prepared the enzyme from erythrocytes of goats, oxen, rabbits, whales and men, thus differentiating its effect from that of hemoglobin 11,13 . The term CA was coined by their co-worker, Philip Eggleton, but his name is not mentioned in the articles 1 . In 1961, after development of the enzyme nomenclature, CA was listed in Enyzme Commission Class 4 (EC 4.2.1.1). Until 1976, isoenzymes I and II were only known, but since then, a total of 18 isoenzymes have been described which belong to distinct genetic families (alpha-, beta-and gamma-CAs). Their functional differences have been described in detail and the structure was elucidated by Roentgen diffraction.

The CA theory of acid secretion
The formation of hydrochloric acid secretion by the stomach was an interesting topic in gastric physiology for decades. In 1934, Franklin Hollander (1899-1966 of Mount Sinai Hospital, New York, formulated the two components theory. Accordingly, gastric juice comprises a parietal cell-secreted acid solution and an alkaline one produced by non-parietal cells 14 . The retrodiffusion theory belongs to Torsten Teorell (1905Teorell ( -1992) (Karolinska Institute, Stockholm). Hydrogen ions are primarily secreted by the parietal cells and are exchanged with NaCl from plasma through the mucosa 15 . In 1932, the so-called alkaline tide was observed (i.e. increase of blood bicarbonate level during stimulation of HCl secretion) and based on this, Davenport presumed the existence of gastric CA, which was subsequently demonstrated in 1939. In 1940, the author formulated the CA theory in which by catalysing the hydration of CO 2 and further dissociation to H + and HCO 3 , the enzyme represents the main source of hydrogen ions 12 . And, for the first time, he also demonstrated the inhibition of CA and that of gastric acid secretion by thiocyanate 16 . His theory did not explain the driving force which moves H + ions against a huge concentration gradient between the gastric lumen and parietal cells. Thus, in 1946, the author himself withdrew his theory 1 .
Ioan Puşcaş also envisaged a central role for CA in the parietal cells. In a series of original experiments, using Wilbur-Anderson's electrometric and colorimetric method (1948) and Maren's simplified micromethod (1960), he showed that mediators of gastric acid secretion (histamine, pentagastrin, acetylcholine and calcium ions) activate gastric mucosa CA depending on the dose 17,18 . The effect of histamine on CA was, however, He also determined a minimal level of pH ¼ 6.5 below which a coma occurs and showed the relation between renal blood flow and urea clearence.

1932-1933
Norman Urquhart Meldrum (1907Meldrum ( -1933 11 19 and confirmed in 1972 by Russian authors 20 . In 1984, the validity of this phenomenon was verified on isolated guinea pig oxyntic cells. The activity of all the secretagogues was inhibited by the new AB Hassle product H149/ 94, suggesting a functional link between CA and the proton pump 21 . In 1999, Ioan Puşcaş et al. 22 showed that omeprazole inhibits both gastric mucosa H + /K + -ATP-ase and CA. Finally, in 2011, Supuran et al. 23 demonstrated the activation of CA I and II by mono-and dihalogenated histamine derivatives at subnanomolar concentrations. Ioan Puşcaş also showed that H 2 histamine receptor blockers (cimetidine and ranitidine), muscarinic antagonist pirenzepine, Zn ions and prostaglandins inhibit gastric mucosa CA. However, the importance of these phenomena in the parietal cell physiology was never shown.

Treatment of gastric and duodenal ulcers by CA inhibitors
Shortly after the discovery of acetazolamide, Henry D. Janowitz (1922-2008) (Mount Sinai Hospital, New York) administered it in dogs 24 and humans 25 . Some years later, it was shown that higher diuretic doses of acetazolamide are needed for several days to ensure the efficient inhibition of acid secretion 26 . Work done in 1956 at the Chemical Defence Experimental Establishment, Liverpool, showed that acetazolamide in usual doses did not abolish HCl secretion 27 , but, in 1956, Rodolfo Cheli (1928-1997 reported some inhibitory effects in normal subjects and chronic gastritis patients 28 . In 1960, acetazolamide was introduced to the therapy of peptic ulcers in a military hospital from Illinois, USA 29 . Acetazolamide was given in a dose of 250 mg q.i.d. to 125 patients for 7-20 days and along with a decrease of acid secretion, the typical ulcer symptoms were relieved promptly. However, no radiographic control was performed and the drug produced reversible metabolic acidosis. In 1963, a German physician working in Ghana 30 administered 3 Â 2 tablets of 250 mg acetazolamide/day to 11 patients for 9 days and also noted a 70-92% decrease in histamine-stimulated acid secretion. In the same time, CA inhibitors were used in several other conditions ( Table 2).
In 1968, Ioan Puşcaş began to use acetazolamide inŞimleu Silvaniei Hospital and he rapidly gained experience with a large number of patients. The results of his studies were largely published in the national and international medical press. In several consecutive studies, he showed that acetazolamide, given in dose of 250-500 mg/t.i.d. for 10-21 days, healed gastric and duodenal ulcer in high percentages, approaching 100%. These results were unparalleled with any other treatment in those days 31,32 . To compensate the electrolytic losses stemming from the diuretic effect of the drug, a mixture of sodium, potassium and magnesium salts was added -without any antacid effect -and the whole composition was marketed as Ulcosilvanil (ulcer + Transylvania). These high-healing percentages were documented radiologically and later by fibre-optic or videoendoscopy. Similar results were obtained with ethoxzolamide 33 . The drug was also efficient as maintenance treatment for preventing peptic ulcer relapses when administered for 10 days/month 34 . However, the high efficacy was overshadowed by a wide spectrum of side effects (fatigue, paraesthesias, renal colic, loss of appetite, vision disturbances, decreased libido and headache) in an era of rapid development of clinical pharmacology, when the ''safety'' of any drug was more important than its ''efficacy''. Although these side-effects were recognized from the beginning of acetazolamide use, they were often masked or only partially reported in the published studies. Thus, Ulcosilvanil, although marketed countrywide, remained a miracle drug given in its parent town,Şimleu Silvaniei, for more than two decades. The advent of the safer histamine H 2 receptor blockers, proton pump inhibitors and the discovery of Helicobacter pylori seriously curtailed the use of Ulcosilvanil and because of a drop in sales, its production was terminated before the end of the last century. Ioan Puşcaş was aware of the discovery of H. pylori, but he never recognized the importance of the bacterium in peptic ulcer disease (see Appendix and Supplementary Table).
Helicobacter pylori CAs and their inhibition: the rationale of treating peptic ulcers with CA inhibitors As other bacteria, H. pylori encodes for two CAs, an a-class and a b-class enzyme, called HpaCA and HpbCA, respectively 35,36 . The two enzymes possess a significant CO 2 hydrase activity, as determined by several groups, which is probably due to their crucial physiological roles for the bacterium 37,[38][39][40][41][42][43][44] .
HpaCA is composed of 247 amino acid residues and shows 27-36% similarity with other a-class bacterial CAs, including Klebsiella pneumonia, Neisseria gonorrhoeae, Enterococcus faecalis, Anabaena PCC7120 and Synechococcus PCC7942 37,38,45 . Surprisingly, the HpaCA amino acid sequence shows relatively high similarity (23-36%) with the 15 types of human CAs (hCA), although there is a large evolutionary distance between the two species, indicating thus that CAs possess a fundamental biological function which has been conserved during the evolution of life on earth. Most amino acid residues which are known to be involved in the catalytic mechanism or to form a network of hydrogen bonds are conserved between hCAs and HpaCA, including the three zinc-liganded His residues (at position 94, 96 and 119, hCA I numbering system) and several important residues for the binding of the substrates, inhibitors and activators, such as Tyr7, His64 and Leu198 [45][46][47][48] .
HpbCA is composed of 221 amino acid residues and shows 25-34% similarity with other b-class bacterial CAs, including the two enzymes from Escherichia coli, the one from Synechococcus elongatus, Brucella suis and Haemophilus influenzae 35,36,49,50 . It has been reported that most of the bacterial b-CAs are composed of three sequential components: (i) an N-terminal arm including two a-helixes (H1 and H2), (ii) a zinc-binding core including three b-sheets (S1-S3) and two a-helixes (H3 and H4) and (iii) a C-terminal subdomain including two b-sheets (S4-S5) and five a-helixes (H5-H9) 51,52 . The amino acid sequence of the zincbinding core of HpbCA is highly conserved among other bacterial b-CAs sequenced to date. These residues include the Zn(II)coordinating amino acids Cys42, Asp44, His98 and Cys101 (residue numbering are based on the E. coli CynT2 numbering system) 51,52 . The principal difference between aand b-CAs consists in the fact that generally b-CAs are oligomers usually consisting of two to six monomers of molecular weight of 25-30 kDa 51,52 .
The catalytic activity of recombinant, purified HpaCA and HpbCA for the physiologic reaction (CO 2 hydration), in comparison with that of several a-CAs of human origin, such as hCA I-III (cytosolic isozymes), hCA VA and VB (mitochondrial isoforms) and hCA XII and XIV (transmembrane isozymes), are shown in Table 3. It may be observed that HpaCA and HpbCA are catalytically efficient CAs, possessing a high enzymatic activity (the b-class enzyme is 3.2 times more active than the a-CA from this bacterium) [38][39][40][41] . Furthermore, this activity is almost identical (as k cat /K m value) to that of hCA I, whereas the K m value of the bacterial enzyme is closer to that of hCA II than that of hCA I. In fact, HpbCA is a medium-efficiency CA, possessing a catalytic activity higher than that of hCA III, hCA VA, hCA XII and hCA XIV among others. Only hCA VB and especially hCA II, one of the best catalysts known in nature, show a better activity than HpbCA [38][39][40][41] . Also, it may be observed that the activity of all these enzymes is inhibited by the CA inhibitor (CAI) par excellence, the sulfonamide drug, acetazolamide AAZ [53][54][55][56][57] , which may explain why Ulcosilvanil showed a range of severe side effects, due to inhibition of CA isoforms in other organs than the stomach, or better to say, due to hCAs and not Hpa/bCA inhibition (Table 3). Chirica et al. 37 studied the subcellular localization of HpaCA and HpbCA by using electron microscopy with immunonegative staining and SDS-digested freeze-fracture immunogold labelling and then showed that hpaCA was attached to the outer membrane and that hpbCA was localized in the cytosol, on the cytosolic side of the inner membrane, and on the outer membrane facing the periplasmic space 37 .
What are the biological roles of periplasmic HpaCA and cytosolic HpbCA in H. pylori? Stähler et al. 58 reported that HpaCA and HhpbCA deletion mutants as well as the double mutant displayed a significant decrease in urease activity, indicating a metabolic link of these two enzyme types, urease and CAs. They also showed that only the HpbCA deletion mutant showed a clearly reduced growth at pH 6.0-6.25 for 48 h in culture, compared to the parental strain, the HpaCA deletion mutant and the double mutant, suggesting thus that the cytosolic HpbCA is only important for acid resistance when the periplasmic HpaCA is functional 58 . In another study by Marcus et al., 59 the HpaCA deletion mutant showed an $3 log 10 decrease in survival after 30 min of exposure to pH 2.0 when compared with the wildtype strain. These discrepant findings are probably due to the different culture conditions employed in the two studies. In the presence of the CAI acetazolamide (AAZ), acting against both hpaCA (K I : 21 nM) and hpbCA (K I : 40 nM) (Table 3), another H. pylori strain also (ATCC 43504) showed an $3 log 10 decrease in survive 59 . These findings indicate that both hpaCA and HhpbCA might be essential for the bacterial survival.
Marcus et al. 59 has proved that HpaCA is essential for the acid acclimation and survival of the pathogen and then proposed a model for the role of urease and hpaCA in maintenance of the periplasmic pH in acidic environment 58,59 . Marcus et al. 59 employed a CA deletion mutant of H. pylori and showed that the generation of NH 3 by urease and bicarbonate by hpaCA has a major role in the regulation of the periplasmic pH 6.1 and inner membrane potential À101 mV under acidic conditions 59 . Thus, buffering of the periplasm to a pH consistent with viability depends not only on the ammonia efflux from the cytoplasm but also on the conversion of CO 2 (produced by urease) to bicarbonate by the periplasmic hpaCA (Figure 1). In fact, the pKa of the carbonic acid/bicarbonate couple of $6.1 is very appropriate for such a task, unlike the ammonium/ammonia buffer, which having a pKa of 9.2, is less useful for buffering the periplasm to pH values close to neutrality. The similar reactions to the one mentioned above for the periplasm will occur in the cytoplasm, i.e. hydration of CO 2 by cytoplasmic hpbCA producing bicarbonate to buffer the cytoplasmic pH at $8.0 (Figure 1) 59,60 . Another product of the hpbCA-catalysed reaction, i.e. the protons, are neutralized by NH 3 , the product of urease. NH 3 , can also neutralize entering protons, which may occur in the highly acidic condition present in the stomach. Accordingly, in cooperation Table 3. Kinetic parameters for CO 2 hydration reaction catalysed by some human a-CA isozymes at 20 C and pH 7.5, and HpCA isozymes belonging to the aand b-CA class, and their inhibition data with acetazolamide AAZ (5-acetamido-1,3,4-thiadiazole-2-sulfonamide), a clinically used sulfonamide [39][40][41]   with cytoplasmic urease, hpaCA and HpbCA work for acid acclimation of H. pylori in its periplasm and cytoplasm, respectively, suggesting that these CAs may be attractive drug targets for developing anti-H. pylori agents. Preliminary in vitro studies, with the potent and clinically used CAI, AAZ (K I : 21 nM for HpaCA and 40 nM for HpbCA) 41 from Lindskog's group 37 did not show any inhibition of bacterial growth at the concentration of 5 mM. However, in the presence of 1 mM of AAZ, there was an $3 log 10 decrease in acid survival of H. pylori 37 . Nishimori's group 41 also published data showing that another CAI, methazolamide (MZA), displayed growth inhibition of certain strains of H. pylori in vitro. MZA possesses weaker inhibitory activity than AAZ but it is however a strong inhibitor of both HpaCA and HpbCA. Interestingly, the growth of the bacterial strain SS1 was totally inhibited at 100 mg/ml of MZA, but there was no inhibitory effect of the drug at this concentration on the growth of another strain 41 , 11 637. However, at a concentration of 500 mg/ml of MZA, the growth of strain 11 637 was also inhibited. These findings indicate that CAIs suppress the growth of H. pylori, but the susceptibility to the drug greatly varies in each strain 41 .

Conclusions
The two CAs present in H. pylori together with the urease play a central role for acid acclimation and survival in the highly acid condition of the stomach of this pathogen. Several sulfonamide CAIs have been shown to possess inhibitory effect on the growth of H. pylori in vitro, as well as inhibition of acid secretion from the parietal cells, as shown by Ioan Puşcaş 32 . These findings strongly recommend the clinical use of CAIs as novel agents for the eradication therapy of H. pylori. This may include, of course, combination therapy of a CAI with various antibiotics. Fortunately, no serious complication of AAZ were observed so far in previous clinical studies, apart for the appearance of mild paraesthesias in the limb and around the mouth, which were frequently reported 32,33,61 . For the development of alternative eradication therapies that possess different pharmacological mechanism from those of the previously used drugs, clinical trials with a carefully designed protocol using a panel of CAIs should be warranted. Furthermore, the recent report 62 of the X-ray crystal structure of HpaCA complexed with acetazolamide and methazolamide might lead to the development of bacterial CAselective inhibitors with less affinity for the human enzymes, which might lead to anti-ulcer agents with reduced side effects when compared with AAZ. In conclusion, although the starting hypothesis for treating peptic ulcers with sulfonamide CAIs was wrong, the treatment was efficacious and this was due to inhibition of H. pylori CAs, which are involved in crucial physiologic processes in the bacterium. Finding CAIs selective for the bacterial over the human enzymes may lead to efficient treatment approaches for this disease, which was in a way pioneered, but not thoroughly explored by Ioan Puşcaş and his collaborators. paediatrics, also realized in a period of marked economic decay. Journalists claim he authored more than 600 publications: most of them, however, are congress abstracts, as presented in Table 3. He and his co-workers wrote some books on peptic ulcer and CA, published in Romania and book chapters published in Hungary, USA and Brazil. In 1985, he organized an international symposium of ''Progress in the Pathophysiology and Treatment of Gastric and Duodenal Ulcers'', attended by leading specialists from all over the world. In 1987, he was appointed director of the new hospital and in 1990, leader of the Centre for Healthcare and Research. He was also professor of gastroenterology at the Faculty of Medicine of the newly established University Vasile Goldiş in Oradea, teaching doctoral fellows. He had 42 patents (drug compositions and diagnostic procedures), but apart from Ulcosilvanil, no one was realized and marketed as product. Doctor Honoris Causa and a member of the United States Academy of Sciences, the hospitalŞimleu Silvaniei took his name in his lifetime. In 1988, he was successfully operated on in Vienna for malignant prostate tumour. He died on 4 April 2015 in his hospital, after a short course of renal failure. He married twice and had two children. His daughter, Carmen, continues his work and one of his four nephews has recently started his medical studies.
Supplementary material available online.
Supplementary Table.