|Home | About | Journals | Submit | Contact Us | Français|
Carbonic anhydrases (CA) catalyze the inter-conversion of CO2 with HCO3 and H+, and are involved in a wide variety of physiologic processes such as anion transport, pH regulation, and water balance. In mammals there are sixteen members of the classical α-type CA family, while the simple genetic model organism C. elegans codes for six αCA isoforms (cah-1 through cah-6).
Fluorescent reporter constructs were used to analyze gene promoter usage, splice variation, and protein localization in transgenic worms. Catalytic activity of recombinant CA proteins was assessed using Hanssons histochemistry. CA’s ability to regulate pH as a function of CO2 and HCO3 was measured using dynamic fluorescent imaging of genetically-targeted biosensors.
Each of the six CA genes was found to be expressed in a distinct repertoire of cell types. Surprisingly, worms also expressed a catalytically-active CA splice variant, cah-4a, in which an alternative first exon targeted the protein to the nucleus. Cah-4a expression was restricted mainly to the nervous system, where it was found in nearly all neurons, and recombinant CAH-4A protein could regulate pH in the nucleus.
In addition to establishing C. elegans as a platform for studying αCA function, this is the first example of a nuclear-targeted αCA in any organism to date.
A classical αCA isoform is targeted exclusively to the nucleus where its activity may impact nuclear physiologic and pathophysiologic responses.
Carbonic anhydrases (CA, E.C. 184.108.40.206) are zinc-containing metalloenzymes (except for the ζ form, which utilizes cadmium instead) that catalyze the reversible hydration of carbon dioxide (CO2) to bicarbonate ions (HCO3−) and protons (H+). CA’s are divided into several distinct classes (α, β (which likely includes the class previously categorized as ε), γ, δ and ζ), of which mammalian CAs belong to the α-class. To date, sixteen CA genes have been identified in humans, with isozymes distributed between the cytoplasm (CAI, II, III, VII, and XIII), the cell membrane (CAIV, IX, XII, and XIV), the mitochondria (CAVa and CAVb), and the extracellular space (CAVI) [1–2]. The CAXV gene is expressed in rodents, but appears to have become a pseudogene in primates . Although most of the human CAs are catalytic, at least three of them (CAVIII, CAX, and CAXI) are not. Catalytic CA enzymes have two conserved features, a Zn2+ ion linked to a histidine triad through imidazoles  and a fourth histidine acting as a proton shuttle . Gene products that are acatalytic appear to lack one or more of these features.
In animals, CAs participate in pH homeostasis, CO2 and bicarbonate transport, water and electrolyte balance, and biosynthetic reactions; the precise role of individual CA isozymes is determined by their cellular and tissue expression, subcellular localization, and catalytic rate . For example, mammalian CAII is an intracellular protein that is widely expressed, it has a fast rate of catalysis that has been deemed diffusion-limited, and its role in transport metabolons that link CA activity to HCO3− transporters has been suggested to maximize HCO3− membrane transport processes [6–10]. As such, it is not surprising that mutations in CAII have been linked to a variety of disorders including osteopetrosis, renal tubular acidosis and cerebral calcification . In contrast, the expression of the GPI-anchored CAIV isozyme on the cell surface is thought to help buffer the extracellular space in the brains of mice  and CAIV has been shown to interact functionally with AE3 to mediate Cl−/HCO3− exchange . It is possible that this regulation of extracellular pH may contribute to synaptic excitability [14–15]. However, it is equally possible that a non-catalytic function of CAIV may contribute to its physiologic role in select cells, as recent data have suggested that the transport activity of the high-affinity monocarboxylate transporter MCT2 is enhanced by CAIV independent of the latter’s catalytic activity . Though CAIV has been shown to be expressed in a variety of tissue types, thus far CAIV mutational phenotypes have been limited to the retina .
In addition to their normal physiologic functions, alterations in CA expression have been correlated with pathologic conditions. For example, CAIX is regulated by hypoxia-inducible factor, hypoxia is a hallmark of many cancers, and CAIX has been shown to be elevated in many tumor types [18–19]. In fact, CAIX antibodies are useful as a diagnostic marker for tumors [20–23] and CAIX inhibition is currently being explored as a therapeutic strategy for cancer . Similarly, acetazolamide, a classic CA inhibitor, is used to treat various diseases/disorders, including glaucoma , epileptic seizures , brain swelling following surgery , and altitude sickness .
Although CAs are widely studied, our understanding of them remains veiled. One of the most obvious gaps is a lack of knowledge as to the normal physiologic role of the acatalytic isoforms. The genome of the model organism C. elegans codes for six α-CA isoforms (cah-1, 2, 3, 4, 5 and 6), and three of these lack the necessary components for catalytic activity (cah-1, 2 and 6). Thus, C. elegans may be a useful reductionist model for discerning the function of both catalytic CAs as well as the evolutionarily conserved acatalytic proteins. Previously, the temporal expression patterns of the worm CA gene family has been studied  and one of these genes, cah-4, has been shown to be regulated by both environmental pH  and oxygen levels . Cah-4 may also play a role in the progression of muscle degeneration in a worm model of muscular dystrophy .
Here, we experimentally defined the transcripts arising from the six cah genes in worms, including their usage of alternative promoters and splice variation. We generated transgenic nematodes to determine the gene expression patterns and intracellular location of the six worm cah gene products, as well as how splice variation affected protein targeting. In addition, we expressed the C. elegans cah cDNAs and their splice variants in mammalian tissue culture cells, then assessed protein localization and activity, respectively.
To our surprise, we found a splice variant of cah-4, the same gene implicated by three separate groups as being physiologically important [30–32], that was targeted to the cell nucleus. Nuclear expression, which required a 45 amino acid N-terminal extension coded for by an alternative first exon, occurred nearly exclusively and ubiquitously throughout the nervous system. Though an RNA splicing factor has previously been shown to have catalytic CA activity , our results are the first to demonstrate that a classical αCA resides in the nucleus of any organism. Both of the cah-4 splice variants were found to be catalytically active, and we hypothesize that CAH-4A may contribute to nuclear pH regulation or oxidative stress resistance.
In conclusion, the results presented here form a foundation upon which to structure genetic approaches using a powerful model system to identify new functions for a widespread, evolutionarily-conserved gene family
Nematodes were routinely cultured at either 16°C or 20°C on Normal Growth Media agar plates seeded with OP50 bacteria. To create transgenic lines, young adult worms (generally pha-1(e2123ts)III mutants) grown at the permissive temperature of 16°C were microinjected with DNA at a final concentration of 150 ng/μl in high-potassium injection buffer . In general, equal amounts of experimental DNA and rescue marker (pCL1, a vector coding for pha-1) were co-injected. In some cases, over-expression of GFP-tagged CA was toxic to worms, and in these cases the amount of experimental DNA in the injection mix was reduced.
The strains developed and reported on in this work are:
Transcriptional fusions (referred to in the figures as Pgenename::GFP) — KWN333, pha-1(e2123ts)III rnyEx199 [Pcah-1::GFP PCR; pCL1 (pha-1+)]; KWN334, pha-1(e2123ts)III rnyEx200 [Pcah-2a::GFP PCR; pCL1 (pha-1+)]; KWN335, pha-1(e2123ts)III rnyEx201 [Pcah-2b::GFP PCR; pCL1 (pha-1+)]; KWN336, pha-1(e2123ts)III rnyEx202 [Pcah-3::GFP PCR; pCL1 (pha-1+)]; KWN337, pha-1(e2123ts)III rnyEx203 [Pcah-5::GFP PCR; pCL1 (pha-1+)]; KWN338, pha-1(e2123ts)III rnyEx204 [Pcah-6::GFP PCR; pCL1 (pha-1+)].
Translational fusions (referred to in the figures as Pgenename::PROTEIN::GFP) — KWN339, pha-1(e2123ts)III him-5(e1490)V, rnyEx205 [Pcah-5::CAH-5::pHluorin PCR; pCL1 (pha-1+)]; KWN340, pha-1(e2123ts)III him-5(e1490)V, rnyEx206 [Pcah-3::CAH-3::GFP PCR; pCL1 (pha-1+)]; KWN348, pha-1(e2123ts)III him-5(e1490)V, rnyEx213 [Pcah-1::CAH-1::GFP PCR; pCL1 (pha-1+)]; KWN349, pha-1(e2123ts)III him-5(e1490)V, rnyEx211 [Pcah-2b::CAH-2B::GFP PCR; pCL1 (pha-1+)]; KWN350, pha-1(e2123ts)III him-5(e1490)V, rnyEx214 [Pcah-6::CAH-6::GFP PCR; pCL1 (pha-1+)]; KWN351, pha-1(e2123ts)III him-5(e1490)V, rnyEx215 [Pcah-2a::CAH-2A::GFP PCR; pCL1 (pha-1+)].
Cah-4 fusions: KWN172, pha-1(e2123ts)III him-5(e1490)V, rnyEx096 [Pcah-4a::GFP PCR; pCL1 (pha-1+)]; KWN35, pha-1(e2123ts)III him-5(e1490)V rnyEx013 [pIA3-R01a (Pcah-4a::CAH-4A::GFP), pCL1 (pha-1+)]; KWN36, pha-1(e2123ts)III him-5(e1490)V rnyEx014 [pIA3-R01b (Pcah-4b::CAH-4B::GFP), pCL1 (pha-1+)]; KWN377, pha-1(e2123ts)III rnyEx232 [Pcah-4a::GFP::CAH-4A PCR, pCL1 (pha-1+)]; KWN378, pha-1(e2123ts)III rnyEx233 [pKT23 (Pnhx-2::CAH-4A exon 1::GFP), pCL1 (pha-1+)];
Random hexamer primers were used to synthesize cDNA from a mixed-stage population of C. elegans using an iScript Tm cDNA synthesis kit (BioRad, Hercules, CA). Nested gene-specific oligonucleotides were then employed in consecutive rounds of PCR in combination with an adaptor primer for 3′ RACE or SL1/SL2 leader primers for 5′ RACE (almost all C. elegans mRNAs have a non-template encoded 5′ trans-spliced 22 nt leader added post-transcriptionally, with SL2 leaders being good indicators of the ~15% of worm genes that are downstream in an operon). Each individual PCR product was gel-isolated, cloned, and sequenced.
The predicted ORFs for each cah gene product were cloned from gel-isolated PCR fragments into the vector pcDNA3.1-V5/His/topo (Invitrogen, Carlsbad, CA) to create pcDNA3.1cah-1 through pcDNA3.1cah-6. Each insert was fully sequenced on both strands and was engineered to lack the endogenous stop codon. Instead, a V5 epitope was encoded in-frame at the C-terminus of each ORF; this epitope was recognized by a commercially available mouse monoclonal anti-V5 antibody (Invitrogen). CHO cells that had been transiently transfected with the cah expression vectors were fixed in PBS/2% paraformaldehyde/50% methanol for 20 min. at 4°C, then permeabilized and blocked in PBS/0.1% Triton X-100/5% BSA/1% Normal Goat Serum for 1 hour at RT prior to antibody incubation (1:2000 dilution in blocking buffer). Following 3x washes in PBS/0.1% Triton X-100, the anti-V5 antibody was detected using a 1:5000 dilution of rabbit anti-mouse IgG conjugated to Alexa Fluor 488.
The rabbit CAIV expression vector was a kind gift of Dr. George Schwartz (Univ. of Rochester).
In order to create transcriptional and translational fusions of the cah promoters and genomic ORFs to GFP for expression in transgenic worms, PCR sewing was used. In short, a PCR product amplified from C. elegans genomic DNA was engineered with a non-template encoded 5′ extension built into one of the oligonucleotides such that it was complementary to the 5′ end of a separate PCR product coding for GFP (with the unc-54 3′ UTR added for mRNA stability). The two PCR products were gel-isolated, combined, and allowed to progress through five rounds of PCR amplification in the absence of primers, resulting in the annealing and extension of the two products. Prior to the sixth round of amplification, a 5′ nested primer targeted at the promoter::ORF product and a 3′ nested primer targeted at the unc-54 3′ UTR product were added and the reaction was allowed to progress for an additional 20 cycles. The final full length product was phenol/chloroform extracted, precipitated with ethanol, and used for microinjection. The sizes of the promoter fragments used to drive expression in the PCR fusions were: cah-1, 3470 nt; cah-2a, 1399 nt; cah-2b, 1610 nt; cah-3, 3319 nt; cah-4a, 348 nt; cah-4b, 3286 nt; cah-5, 3056 nt; cah-6, 3531 nt.
Standard molecular cloning techniques were used. The vector pKT23 (Pnhx-2::CAH-4A EXON 1::GFP) was created by PCR amplification of the 45 amino acid coding region found in the first exon of cah-4a using Acc65I-tagged primers, followed by digestion and cloning into the Acc65I site of pIA5-nhx-2 . The vectors pIA3-RO1a and pIA3-RO1b were likewise created by PCR, using C. elegans genomic DNA as a template. pIA3-RO1a contains a region from immediately downstream of the first (non-coding) exon in cah-4b to the end of the cah-4 coding region, cloned in frame with GFP in the base vector pIA3  as an NheI-SacII fragment. pIA3-R01b contains the cah-4b promoter cloned as an NheI-SacII fragment into pIA3, with the cah-4b coding region then cloned as a SacII fragment into the resulting vector. This effectively removed the cah-4a promoter and first exon from the construct.
pKT18 was created by PCR cloning the ratiometric fluorescent biosensor pHluorin cDNA  amplified using a primer containing a 5′ non-template encoded SV40 NLS into the topo TA cloning vector pcDNA3.1 V5/His/topo (Invitrogen Corp., Carlsbad, CA). The insert was fully sequence on both strands.
pKT18 was transiently transfected into CHO cells in culture using Lipofectamine 2000 (Invitroge., Carlsbad, CA) as recommended by the manufacturer. Immediately following transfection, cells were trypsinized onto glass coverslips and allowed to adhere under normal growth conditions overnight. The following day, a coverslip containing adherent cells was placed into a perfusion chamber residing on the stage of a Nikon Eclipse 2000 inverted microscope (Nikon Instruments Inc., Melville, NY) equipped with a monochromator (TILL Photonics, Germany) and Cooke sensicam (Cooke Corp., Romuslus, MI) running TILLvisION software for image acquisition. The cells were superfused with buffer containing HCO3− (in mM): 115 NaCl, 20 NaHCO3−, 5.4 KCl, 0.4 KH2PO4, 0.33 NaH2PO4, 10 glucose, 20 Hepes, 1.2 CaCl2, 0.8 MgSO4, bubbled with 5% CO2 and adjusted to pH 7.4 with Tris. Following equilibration, the cells were switched to a HCO3− free buffer (in mM): 135 NaCl, 5.4 KCl, 0.4 KH2PO4, 0.33 NaH2PO4, 10 glucose, 20 Hepes, 1.2 CaCl2, 0.8 MgSO4, pH adjusted to 7.4 with Tris base. During this period, 535-nM emissions were measured following excitation at either 410-nm or 470-nm, and the ratio of these emissions was converted to pH by in situ calibration using the high K+/nigericin technique .
CHO cells were plated onto glass coverslips and transiently transfected with pcDNA3.1-based mammalian expression vectors coding for V5 epitope-tagged CAH proteins using Lipofectamine 2000 (Invitrogen). 24-hours post-transfection, the cells were fixed and stained for CA activity using a cobalt-phosphate detection method . Briefly, the cells were incubated overnight at 4°C in 3% glutaraldehyde in 0.1M sodium phosphate buffer pH 7.3 to fix. Following three rinses with 0.1M sodium phosphate buffer (pH 7.3), the cells were incubated for 5 min. in a solution made by combining 17 ml of solution A (containing 1 ml of 0.1M CoSO4, 6 ml of 0.5M H2SO4 and 10 ml of 0.066M KH2PO4) with 40 ml solution B (0.75 g of NaHCO3 in 40ml of ddH2O). Each of these solutions was prepared fresh from stocks as indicated. The cells were then rinsed in 0.1M sodium phosphate buffer (pH 7.3), and incubated in freshly-made 0.5% (NH4)2S for 2 minutes. The reaction was halted by rinsing with ddH2O. To control for non-enzymatic CA activity, the cells were pre-incubated in 10−4M acetazolamide for 2 hours.
An in silico analysis based upon conserved regions from the mammalian αCA family indicated that the C. elegans genome contains the coding potential for six αCA isoforms. 5′ and 3′ Rapid Amplification of cDNA Ends (RACE) was used to experimentally determine the products arising from transcription of these six putative genes. Two rounds of RACE were performed using nested gene-specific primers designed based upon the predicted DNA sequences in well-conserved regions of the putative ORFs. 3′ RACE products were amplified using an adaptor primer that was tagged onto the dT18 primer used during cDNA synthesis, while 5′ RACE products were amplified using upstream primers corresponding to either SL1 or SL2 trans-spliced leader sequences, which are 22 nucleotide RNAs that are added post-transcriptionally to most genes in worms. All of the major products arising from these reactions were purified, cloned, sequenced and aligned to a genomic map to determine transcriptional start sites and splice variation.
A schematic of our results, including the exons (shown in blue), 5′ and 3′ UTR regions (shown in gray), and leader sequences (SL1) is drawn to scale (in kb) in Figure 1A. As indicated, each of the six cah gene products was found to be trans-spliced to an SL1 leader, and there was no evidence of SL2 splicing (data not shown). SL2 leaders are generally restricted to the <15% of transcripts in worms that occur in operons. As is often the case with worm transcripts, each of the genes contained very short 5′ UTRs, while the 3′ UTRs were of varying length. In addition, two splice variants were identified for the cah-2 and cah-4 genes. This variation occurred at the 5′ ends and resulted in alternative first exons. For both cah-2 and cah-4, transcription of the downstream splice variant was driven by a promoter contained within the first intron of the upstream splice variant, as indicated (Figure 1A).
Protein sequences were determined by virtual translation of the cah ORFs, which were then aligned to the CA isozymes from mammals. The slanted cladogram in Figure 1B indicates the relatedness between the six cah αCAs from C. elegans and CAI through CAXIV from humans (with CAXV being a pseudogene). Interestingly, the CAs cluster into three distinct groups. These include soluble isozymes, membrane-associated isozymes, and acatalytic isozymes. Cah-1, cah-2, and cah-6 are found in a clade with the acatalytic isozymes, which is not unexpected given their lack of certain residues necessary for catalytic activity. Cah-3 and cah-5 are most closely associated with the soluble, catalytic human isozymes. Cah-4 is more closely related to the catalytic isozymes. Not surprisingly given the expansion of the CA gene family in mammals, the cladogram in Figure 1B does not predict direct orthologs and the closest relationship between the C. elegans and human CAs are found amongst the acatalytic isozymes. There is little known as to the molecular function of this sub-class of CA isozyme in mammals, but the intriguing observation that half of the αCAs are presumably acatalytic in C elegans suggests an evolutionarily conserved and potentially important biologic purpose.
To better understand the physiologic role of each of the six cah gene products in worms, we determined where they were expressed. Genomic PCR products containing the cah promoter regions were used to drive the transcription of GFP in transgenic worms, and GFP expression was assessed by fluorescent microscopy. The genomic regions used as promoters are denoted by arrows in Figure 1A.
Neuronal expression was observed in all transgenic lines (Figure 2). In most cases only subsets of the 302 adult neurons expressed GFP, but in at least two cases, cah-6 (Figure 2) and cah-4a (Figure 3D), expression appeared to occur throughout the entire nervous system. In addition to neuronal labeling, strong expression was observed in the intestine (Figure 2; cah-2a, cah-3, and cah-5), the hypodermis (Figures 2 and and3E;3E; cah-4b and cah-5), and various muscle cells including the vulva and pharynx (Figures 2 and and3;3; cah-2a, cah-3, cah-4a, cah-4b, and cah-5). A synopsis of these expression patterns can be found in Table 1. This list is not inclusive, but does detail the major cell types and organs where expression of each isoform was observed. Two caveats are that promoter-driven GFP expression from transgenic extra-chromosomal arrays only approximates endogenous gene expression, and the expression of genes from arrays is suppressed in the C. elegans germline.
In order to establish the intracellular residence of each CA isozyme, the promoters and ORFs were fused to GFP in lieu of the endogenous stop codon by PCR sewing, and the PCR products were injected into worms to create transgenic lines. Fluorescent images of these lines confirmed the cellular expression profiles established above, and helped to clarify protein localization (Figure S1; Table 1). One important caveat here is that it is unclear whether these GFP fusion proteins are functional, and as with all transgenic and fusion protein analysis, the results need to be interpreted in this context.
Most of the isozymes that were predicted to be catalytic were found in the cytoplasm (Figure S1 and Figure 3; cah-3, cah-4b, and cah-5). Two of the putative acatalytic isozymes were intracellular but punctate (Figure S1; cah-1 and cah-2). These punca did not co-localize with the mitochondrial label MitoTracker Red CMXRos and reducing the concentration of the transgenes in the injection mix by 1:30 gave similar results, which led us to believe that they are not merely due to non-specific protein aggregation (data not shown) The final acatalytic isozyme was associated with areas where neurons make contact with other cells, including synapses and neuromuscular junctions (Figure S1; cah-6). We found no evidence for a mitochondrial CA such as mammalian CAV, nor did we detect either extracellular or membrane-bound CA in worms. While this is not conclusive proof that such an isozyme does not exist, it is suggestive.
The most unexpected finding was the discovery of a nuclear CA isozyme (Figure 3; cah-4a). The cah-4 gene contains two alternative first exons whose expression is driven by distinct promoters (Figure 1). The only difference between the resulting proteins is a 45 amino acid region immediately following the initiator Met in CAH-4A that is absent in CAH-4B. While a cah-4b translational fusion protein was expressed most strongly in the excretory cell, GFP labeling was also found in the cytoplasm of various muscle and hypodermal cells (Figure 3B). In contrast, the cah-4a promoter was expressed throughout the nervous system as well as in the head muscle cells (Figure 3D), and a translational CAH-4A::GFP fusion protein was apparently directed to the nucleus of these cells (Figure 3F). In head muscle cells, unlike neurons, the nuclei are small compared to the overall cell size, and nuclear targeting was readily observable. We also confirmed that GFP fluorescence in neurons co-localized with DAPI staining in the nucleus (Figure S2).
Moving GFP from the C-terminus to the N-terminus of CAH-4A demonstrated that exon 1 was not strictly an N-terminal sorting signal and that it could drive nuclear localization from a site within the fusion protein (Figure 4A). In fact, 45 amino acids coded for by exon 1 formed a bona fide NLS which on its own was sufficient to direct GFP to the nucleus (Figure 4B). This targeting was not limited to neurons, either, as an exon 1::GFP fusion that was driven by the nhx-2 promoter was found in the nucleus of intestinal cells (Figure 4C). Thus, we conclude that CAH-4A is a nuclear CA, and we hypothesize that it contributes to neuronal function through a previously-unrecognized role in nuclear physiology.
To test for catalytic activity of the worm CAs, their ORFs were cloned and expressed as recombinant V5 epitope-tagged proteins in CHO cells. Anti-V5 antibody staining was used to assess recombinant protein targeting, while Hansson’s histochemistry was used to assay catalytic activity. Hansson’s stain is a cobalt precipitate that results from CA activity and serves as a visual indicator of both catalytic potential and the location of catalysis .
As a positive control, cells were transfected with rabbit CAIV, a GPI-anchored extracellular CA that stains at cell-cell contacts when reacted with Hansson’s (Figure 5D). As expected, despite being anti-V5 immuno-reactive, cells transfected with cah-1, cah-2 or cah-6 cDNAs were not stained by Hansson’s technique (data not shown). However, cells transfected with cah-3, cah-4a or cah-4b cDNA were robustly stained (Figure 5E–G). In the case of cah-4a, the staining was predominantly nuclear (Figure 5G). This result also confirmed previous data indicating that the CAH-4B isozyme was catalytic . Each of these three CA isozymes’ activities was inhibited by preincubation with 10 μM acetazolamide (data not shown). Interestingly, we found that CAH-5, though expressed in the cytoplasm, did not result in discernable CA activity using Hansson’s technique (data not shown). This isozyme may have very low catalytic activity or specific requirements that are not met in this assay.
Cells that expressed either cah-3 or cah-4b cDNA were anti-V5 antibody reactive in both the cytoplasm and the nucleus (Figure 5A and 5B), while cells that expressed cah-4a cDNA were reactive strictly in the nucleus (Figure 5C). We attribute the partial nuclear distribution of the CAH-3 and CAH-4B proteins to the fact they are quite small (~30 kDa) and may passively diffuse through the nuclear pore, as well as the fact that the nuclei in these cells are thicker than the cytoplasm, leading to what can appear to be nuclear enrichment when viewed by epifluorescent microscopy. GFP fusion proteins such as shown in Figure 2 are likely beyond the molecular weight range within which free diffusion might occur. Even so, the results shown in Figure 4 are consistent with active targeting of CAH-4A to the nucleus, as very little cytoplasmic signal was observed with either immuno-detection or Hansson’s staining.
CAs can contribute to pH regulation both inside and outside of the cell. We next tested whether CAH-4A, consistent with its catalytic potential, could contribute to pH regulation in the nucleus. Nuclear pH was measured using dynamic fluorescent imaging of the pH-sensitive biosensor pHluorin, which was targeted to the nucleus via an SV40 NLS. Cells were allowed to equilibrate in 20 mM HCO3−- 5% CO2-buffered solution. Following equilibration, washout of CO2 was brought about by switching perfusion to nominally HCO3−- CO2-free solution. This caused a rapid alkalinization whose rate was dependent upon the conversion of HCO3− to CO2. Consistent with the results of Hansson’s staining, CHO cells expressing CAH-4A exhibited an increased rate of alkalinization compared to control cells following switchover (Fig. 6A). In addition, these cells exhibited an elevated resting nuclear pH. Acetazolamide reduced the rate of pH change in CAH-4A expressing cells to a value comparable to that of control cells, but had little effect on the rate in control cells themselves (Fig. 6B). We also found that expression of CAH-4B could accelerate the rate of nuclear pH change, though to a lesser extent than CAH-4A, consistent with its partial nuclear distribution in cultured cells (data not shown). To what extent this is an overexpression artifact is currently unknown. These results highlight the potential of a nuclear CA to contribute to organelle pH regulation. This ability may be particularly relevant in response to changes in HCO3− - CO2 such as might result from cell metabolism or ischemic stress conditions.
The mammalian CAs form a large family of zinc-containing metalloezymes, and individual CAs have been shown to participate in pH regulation, HCO3− and ion transport, water and electrolyte balance, and photosynthesis and respiration [2, 39–40]. As might be predicted, CA gene expression and protein distribution contributes to their particular physiologic roles. For example, mitochondrial CAV in the liver fuels pyruvate carboxylase and carbamoyl-phosphate synthetase, thus contributing to gluconeogenesis and the urea cycle, respectively [41–42]. Similarly, salivary CAVI, the only secreted isozyme in this enzyme family, is thought to help prevent dental plaque by reducing acidity in the oral cavity [43–44]. Further, membrane-bound CAIV and XIV isozymes contribute to extracellular pH buffering in the central nervous system .
The three acatalytic CA isozymes CAVIII, CAX, and CAXI are quite closely related to their active neighbors, and point mutations in the coding regions have been shown to be sufficient to convert a catalytically inactive protein to one with enzymatic CA activity [45–46]. However, very little is known about their physiologic function (for review, see  ). Several intriguing observations regarding CAVIII have been reported: first, like the catalytically active CA IX and XII isozymes, CAVIII is overexpressed in certain cancers  and its expression can promote cancer cell growth and invasiveness; second, CAVIII is present in purkinje cells, cerebellar nuclei and brainstem ; and third, a Car8 mutant mouse (wdl) has a gait disorder and aberrant synaptic morphology in the cerebellum . Since CAVIII has been identified as a binding partner for the inositol trisphosphate receptor and regulates its affinity for substrate , it is possible that the phenotype is related to calcium signaling in the cell.
Interestingly, in C. elegans three of the six αCA isozymes are predicted to be acatalytic based upon their lacking one or more of the conserved histidines required for Zn2+ binding, and our observations using Hannson’s staining confirm this prediction. This suggests that CAs may have a conserved function independent of their role in CO2 metabolism. Large scale RNAi screens have failed to identify a phenotype associated with the loss of any of these three isoforms, but the acatalytic isozymes are expressed mainly in neurons (Table 1), and neurons are typically refractory to RNAi. A more rigorous approach will involve creating deletion alleles for the acatalytic CA genes. In this regard, a strain containing a cah-1 deletion allele (ok2032) has been developed by the C. elegans Gene Knockout Consortium and is viable as a homozygous null, but is presently uncharacterized.
In general, the cell expression patterns that we observed for the worm CA genes (Figure 2 and Table 1) were similar to those described in a recent report . However, our analysis of protein localization using translational GFP fusions provided some additional details of interest. First, based upon the distribution of the CA isozymes and the lack of an overt mitochondrial leader sequence on any of the catalytic variants, we conclude that worms lack a CAV ortholog. Second, the acatalytic isozyme CAH-6 localized to regions of cell-cell contact, including neuromuscular junctions (Figure S1). While the fluorescent resolution was insufficient to determine whether the signal was associated with vesicles inside the cell or the plasma membrane, the pan-neural expression and specific targeting of this CA to synapses suggests that it may play a role in neurotransmission. Given that CAH-6 is one of the acatalytic isozymes, further dissection of its role in neurons may help to shed light on this under-studied sub-class of CAs. Finally, while it has been recognized that the cah-4 gene codes for two alternate splice variants  whose expression is driven by mutually exclusive promoters (Figure 3), our results demonstrated that this splicing generates a CA isozyme containing a 45 amino acid N-terminal extension that is both necessary and sufficient for CAH-4A protein targeting to the cell nucleus (Figure 4). This is the first example of a CA in any organism that resides in the nucleus, and begs the question of what the physiologic role of a nuclear CA might be.
While cah-4b is expressed in hypodermis, excretory cell and muscle, the cah-4a promoter drives expression throughout the nervous system, as well as head muscle cells (Figure 3). The pan-neural expression suggests a potentially conserved function in these cells. Moreover, we and others have shown that CAH-4 isozymes are catalytically active (Figure 5 and [29–30]). If catalytic activity is important for CAH-4A function, this would suggest that the nucleus is capable of responding to changes in HCO3−/CO2 metabolism. In fact, recombinant CAH-4A exhibited clear nuclear activity when expressed in mammalian tissue culture cells (Figure 5), and was capable of regulating nuclear pH in response to fluctuations in HCO3−/CO2 (Figure 6), as might be predicted for a catalytic CA targeted to the nucleus. It is unclear based solely upon this result however whether CAH-4A plays a role in endogenous nuclear pH regulation and similarly, how closely nuclear pH is tied to cytoplasmic pH. In fact, nuclear pH regulation is a relatively unexplored topic in general. It is likely that interrogating the physiologic role of CAH-4A in C. elegans will provide insight to these questions. In this regard, a genetic deletion of cah-4 has been annotated as being lethal in its homozygous state (cah-4(tm2805)X; C. elegans Gene Knockout Consortium), clearly suggesting an important function for at least one of the two cah-4 splice variants. Furthermore, we have observed that overexpressing CAH-4A in transgenic worms can lead to phenotypic abnormalities, including locomotor defects (data not shown), though the molecular basis for this dominant effect is currently unknown.
Interestingly, cah-4 expression has been shown to be increased by hypoxia through the canonical HIF/VHL signaling pathway . CAs are commonly regarded as cytoprotective enzymes, and HIF responsive genes generally promote survival under hypoxic conditions. For example, both mammalian CAIX and CAXII have been shown to be HIF responsive and are pro-survival factors. In particular, CAIX is highly expressed in multiple cancers, where it is used as a tumor marker. Mechanistically, CAIX is thought to facilitate acid diffusion and acid transport, thus contributing to tumor pH regulation (for review, see . Moreover, a role for endogenous CAIX in neuronal maintenance was demonstrated by morphological analysis showing vacuolar degenerative changes in the brains of Car9−/−mice .
It has also been demonstrated that oxidative protein modification of CA isozymes, such as may occur under hypoxic conditions, is associated with disease pathophysiology. For example, protein carbonylation of CAII correlates with neuronal pathology in Alzheimer’s disease  and carbonylation of CAIII contributes to oxidative deficiencies in muscle .
However, despite these observations of parallel regulatory motifs and the general underlying theme that CAs can modulate hypoxic sensitivity, there are no clear mammalian CAH-4 orthologs. As regards CAH-4A, none of the mammalian CAs has been shown to reside in the cell nucleus, nor do any of them contain a canonical nuclear localization signal. We suggest here that worm neurons are more exposed to environmental insults than mammalian neurons and that endogenous protective mechanisms in worms may have evolved into adaptive responses in mammals. This is consistent with the fact that worms are incredibly hypoxic resistant compared to mammals. Thus in the absence of a canonical nuclear CA in mammals, it is worth considering whether situational CA transport to the nucleus could occur, perhaps in response to stress signaling. Alternatively, stressors might induce the expression of a cryptic promoter or splice variant that encodes a nuclear CA isozyme.
Though not of the classical αCA family, in fact there has been one report of a protein capable of catalytic CA activity purified from the nucleus of mammalian cells. The protein was identified as NonO/p54nrb, and histochemical staining revealed CA activity in rat lymphocytes coincident with p54nrb expression . p54nrb is associated with PSF (polypyrimidine-tract-binding-protein-associated splicing factor) and has been reported to be involved in diverse nuclear processes such as transcription, RNA processing, DNA unwinding and repair, and may contribute to the progression of malignant melanoma . Whether CA activity contributes to any of these functions of p54nrb is currently unknown.
If a nuclear CA were catalytically active, our results suggest that it could contribute to nuclear pH regulation, and we predict that this might influence oxidative stress resistance. As alluded to above, however, there is not much known about nuclear pH regulation. Several reports have suggested that the pH of the nucleus is higher than that of the surrounding cytoplasm [57–58], but how this might occur mechanistically is unclear. Establishing a gradient between the cytoplasm and the nucleus would require either a barrier to the diffusion of protons or an extremely a fast enzyme that has H+ or OH− as a reaction component. CAH-4B was shown to have a Kcat/Km of 5.4×107 M−1s−1 , which while not diffusion-limited is still second only to that of mammalian CAII.
Compartmentalization could also help to establish a nuclear pH microdomain. Although the nuclear pore should permit the free diffusion of small molecules such as electrolytes, there is compelling evidence that localized Ca2+ signaling within the nucleus occurs in response to synaptic activity, and this signaling elicits subsequent protective measures . Protection occurs through altering the neuron’s transcriptional profile  and may involve morphologic changes in the nucleus itself [61–62]. The observed alterations to nuclear geometry are consistent with the idea that the nucleoplasm is restricted into signaling microdomains. Since acid diffusion rates can be buffered by CA activity, microdomains may contribute to the ability of CAH-4A to regulate pH homeostasis in the nucleus.
However, the pK of DNA, while affected by base composition, is generally quite low, and the negative charge on the phosphate backbone of DNA will be largely unaffected by changes in pH in the physiologic range, Thus, it is unlikely that small changes in nuclear pH will have large effects on DNA binding to histones or transcription factors. It is possible that there is an alternative mechanism whereby pH might regulate nuclear function. Though pH is most commonly thought of as providing a metabolic context that shapes enzymatic activities in the cell, there is a growing notion that proton recognition may contribute directly to cell signaling [63–66]. It will be fascinating to determine whether CAH-4A contributes to nuclear pH homeostasis and whether the loss of CAH-4A has functional consequences on neuronal survival.
This work was supported in part by USPHS R01 NS064945 (K.N.) and a UNCF/Merck Postdoctoral Fellowship Award to T.A.M. We also acknowledge the C. elegans Genetics Center for strains.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.