Hypoxia-Inducible Carbonic Anhydrase IX and XII Promote Tumor Cell Growth by Counteracting Acidosis through the Regulation of the Intracellular pH

Acidosis of the tumor microenvironment is typical of a malignant phenotype, particularly in hypoxic tumors. All cells express multiple isoforms of carbonic anhydrase (CA), enzymes catalyzing the reversible hydration of carbon dioxide into bicarbonate and protons. Tumor cells express membranebound CAIX and CAXII that are controlled via the hypoxiainducible factor (HIF). Despite the recognition that tumor expression of HIF-1A and CAIX correlates with poor patient survival, the role of CAIX and CAXII in tumor growth is not fully resolved. To understand the advantage that tumor cells derive from expression of both CAIX and CAXII, we set up experiments to either force or invalidate the expression of these enzymes. In hypoxic LS174Tr tumor cells expressing either one or both CA isoforms, we show that (a) in response to a ‘‘CO2 load,’’ both CAs contribute to extracellular acidification and (b) both contribute to maintain a more alkaline resting intracellular pH (pHi), an action that preserves ATP levels and cell survival in a range of acidic outside pH (6.0–6.8) and low bicarbonate medium. In vivo experiments show that ca9 silencing alone leads to a 40% reduction in xenograft tumor volume with up-regulation of ca12 mRNA levels, whereas invalidation of both CAIX and CAXII gives an impressive 85% reduction. Thus, hypoxiainduced CAIX and CAXII are major tumor prosurvival pHiregulating enzymes, and their combined targeting shows that they hold potential as anticancer targets. [Cancer Res 2009;69(1):358–68]


Introduction
Adaptation of tumor cells to hypoxia and acidosis is a critical driving force in tumor progression and metastasis (1,2).Cancer cells produce a large amount of lactic acid (3), which is generated through glucose metabolism and inefficient vascular clearing, resulting in an acidic microenvironment within many solid tumors (4).Extracellular acidosis represents a threat to cell survival by modifying the intracellular pH (pH i ), wherein a 0.1 pH i variation can disrupt multiple biological functions, including ATP production, protein synthesis, cell proliferation, migration, and apoptosis (5)(6)(7).Because numerous intracellular processes require close regulation of pH i , most mammalian cells, particularly hypoxic tumor cells, have developed key strategies to regulate their pH i .Activation of the hypoxia-inducible factor-1 (HIF-1) in hypoxia plays a major role in regulating pH homeostasis by enhancing expression of membrane located transporters, exchangers, pumps and ecto-enzymes (8).To survive in an acidic environment, the pH i -regulating system of tumor cells actively extrudes acids via the growth factor-activated Na + /H + exchanger 1 (NHE-1; refs.9-12) and the monocarboxylate transporters (MCT1 and MCT4; ref. 13).We showed previously that NHE-1 plays a key role in tumor development particularly for cells producing large amounts of lactic acid (14).In the opposite direction to H + extrusion, HCO 3 À influx through Na + -HCO 3 À cotransporters (NBC) and Cl À /HCO 3 À exchangers (AE) contributes to cytoplasmic alkalinization (15)(16)(17).Carbonic anhydrases (CA), which catalyze the reversible hydration of cell-generated carbon dioxide into protons and bicarbonate ions, have also been proposed to contribute to cellular alkalinization (18)(19)(20).The direction of the reaction is dependent on the form, CO 2 or bicarbonate and protons, that predominates.Mammalian cells express 13 active isoforms of CAs, with a conserved active site and variable levels of activity, and 3 inactive isoforms.They differ in their tissue distribution and cellular localization.The expression of the membraneassociated CAIX and CAXII is tightly controlled by oxygen levels in multiple epithelial tumor types (21)(22)(23), and CAIX has a higher extracellular activity than CAXII (23)(24)(25).CAIX is highly induced in an HIF-1-dependent manner (26) and is constitutively expressed in von Hippel-Lindau (VHL)-defective cells.CAXII is up-regulated in VHL-defective renal tumors and induced in hypoxia in tumor cells, but its dependence on HIF is not well established (22).Whereas tumor expression of HIF-1a and CAIX correlate with poor patient survival (19,27), the significance of CAXII, which lacks the extracellular proteoglycan domain of CAIX implicated in cell adhesion (28)(29)(30), is less obvious (24).CAIX may be functionally linked to the regulation of the tumor pH, because it contributes to extracellular acidification (31) and forced CAIX expression in three-dimensional cultured aggregates influences pH i homeostasis (32).However, direct evidence of the conjugated roles of CAIX and CAXII in pH i regulation in cell lines and in tumor growth is still missing.
In this study, we show that the hypoxia-inducible CAIX and CAXII proteins promote cell survival and growth through pH i maintenance.We conclude that CAIX and CAXII constitute a robust pH i -regulating system able to confer a tumor growth and survival advantage on cells exposed to a hypoxic and acidic microenvironment.Finally, as very often hypothesized, but not shown (23,24), we validate here that CAIX and CAXII constitute two new anticancer therapeutic targets.

Cell Culture and Hypoxic Exposure
Chinese hamster lung CCL39 fibroblasts (American Type Culture Collection) and the CCL39-derived mutant PS120 cells, lacking the amiloride-sensitive Na + /H + exchanger (6), were maintained in DMEM (Sigma) supplemented with 7.5% FCS in a humidified atmosphere of 5% CO 2 , 95% air, or 100% air at 37jC.The colon adenocarcinoma cell line LS174Tr expressing the tetracycline (Tet) repressor was provided by Dr. van de Wetering (33) and maintained in DMEM supplemented with 10% FCS.Other human tumor cell lines were likewise cultured.Incubation in hypoxia at 1% O 2 was carried out at 37jC in 95% humidity and 5% CO 2 /94% N 2 in a sealed anaerobic workstation (Ruskinn).

RNA Extraction and Relative and Absolute Real-Time Quantitative PCR
Total RNA was extracted from cells using the RNA extraction kit (Qiagen) according to the manufacturer's instructions.Total RNA (2 Ag) was added to a 20 AL reverse transcription-PCR reaction using the Omniscript kit (Qiagen).The relative expression level of ca9 and ca12 was quantified by real-time quantitative PCR (qPCR), as reported previously (26).The absolute quantification of ca9 and ca12 mRNA was obtained by absolute real-time quantification.A standard curve was prepared using dilutions of the pTREX-ca9 and pTREX-ca12 vectors from 30 to 3 Â 10 8 copies in triplicate.The cycle number of ca9 or ca12 amplification of each extract was compared with the standard curve obtained respectively with pTREX-ca9 or pTREX-ca12 vectors.

Immunoblotting
Cells were lysed in SDS sample buffer.Proteins (40 Ag) were separated on 7.5% SDS polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (Millipore).Membranes were blotted with the M75 antibody to CAIX (Bayer), Hsp90 (Abcam), a polyclonal antibody to recombinant CAXII (Sigma), and a polyclonal antibody to HIF-1a prepared and validated in our laboratory (34).Immunoreactive bands were detected with a horseradish peroxidase (HRP) anti-mouse or anti-rabbit antibody (Promega) by enhanced chemiluminescence (Amersham Biosciences).

Immunofluorescence
Cells at sparse density were grown on glass coverslips and fixed with 3% paraformaldehyde for 30 min followed by saturation for 30 min in PBS-2% gelatin and PBS-1% bovine serum albumin.Cells were then incubated for 1 h with the CAIX antibody without permeabilization, followed by incubation for 1 h with an anti-mouse Alexa 594-conjugated IgG antibody (Invitrogen).Cells were mounted onto slides with citifluor and analyzed with a Leica microscope (objective, 100Â).

In vitro Determination of CA Activity
Cells incubated in normoxia (PS120-pca9 cells) or hypoxia (to induce CAIX and CAXII expression in LS174Tr cells) were placed on ice in normoxia, scrapped into ice-cold PBS, to obtain intact membrane-bound CAIX and CAXII.The cell suspension was immediately centrifuged and resuspended in a bicarbonate-free medium (Sigma) buffered at outside pH (pH o ) 7.4 with 30 mmol/L HEPES.A 0.1 volume of this cell suspension was added to a 3 mmol/L HEPES-buffered solution (HBS) adjusted to pH o 8.2 before rapid addition to a CO 2 -saturated nonbuffered solution and pH determined over time (microelectrode, Schott Instrument) to monitor the rapid hydration of CO 2 to carbonic acid.For inhibition of the total CA activity, 100 Amol/L acetazolamide (ACTZ; Sigma) was added to the cell suspension in normoxia 15 min before the experiment.

Resting pH i Measurement
[ 14 C]Benzoic acid.The pH i was measured using the technique of distribution of the weak acid [7-14 C]benzoic acid (Amersham Biosciences) in intracellular and extracellular spaces for exponentially growing cells (35).Diisothiocyanatostilbene-2 ¶,2-disulfonic acid (DIDS; 1 mmol/L, Sigma) was used as an inhibitor of HCO 3 À transporters.
BCECF-AM probe.Exponentially growing cells seeded on glass coverslips were incubated for 25 min in bicarbonate-free HBS (pH o 7.4), MES-buffered solution (MBS; pH o 6.6), or bicarbonate-buffered solution (BBS; adjusted to pH o 7.4).The pH-sensitive fluorescent dye BCECF-AM 1 Amol/L (Sigma) was then added for 5 min at room temperature.Cells were then quickly washed with the appropriate HBS, MBS, or BBS solution and transferred to a laminar flow cell chamber perfused with the same solution.Ratiometric measurement of the fluorescence of 50 randomly selected individual cells per coverslip was performed in a workstation (Acquacosmos).The pH i was estimated by an in situ two-point calibration (pH o 6.6-7.6) with perfusion of a high K + buffer solution containing 130 mmol/L KCl, 2 mmol/L CaCl 2 , 1 mmol/ L MgCl 2 , 10 mmol/L glucose in 20 mmol/L HEPES-Tris (or MES-Tris), and 25 Amol/L nigericin to allow pH i to equilibrate with the external pH o .

ATP Determination
Cells were seeded in DMEM and grown at different pH o , as described for DNA synthesis.ATP levels were measured using a Cell Titer Glo kit (Promega) according to the manufacture's instructions.The relative luminescence unit was normalized to the quantity of protein.
Dishes were then returned to 5% CO 2 in a regular media for 7 d before staining with Giemsa (Fluka).

Cell Proliferation in Three Dimensions
To grow spheroids, 1,600 cells were seeded in drops in 20 AL of HCO 3 ) adjusted to a pH o of 6.6 or 7.4, respectively.Just before their transfer to a laminar flow cell chamber, BCECF-AM was added to the cells for 5 min.Cells were then perfused with the respective solutions, and a three-point calibration curve was created with high KCl/nigericin solutions to measure pH i as a function of the fluorescence ratio.Significant differences based on the Student's t test (***, P < 0.001; n = 200 cells for each PS120-pev and PS120-pca9 cells at pH o 6.6, n = 3000 cells for pH o 7.4 with or without HCO 3

À
).In the absence of HCO 3 À , PS120-pca9 cells had a statistically higher resting pH i than PS120-pev cells perfused at either a pH o of 6.6 or 7.4.C, CAIX expression increases pH i over a range of pH o .The pH i was determined in PS120-pev and PS120-pca9 cells as a function of pH o with [ 14 C]benzoic acid.Exponentially growing cells of two independent clones for PS120-pca9 and PS120-pev cells were equilibrated for 15 min in a nominally HCO 3 À Àfree solution with pH o varying from 6.2 to 7.4.Cells were then shifted for 15 min to the same solution containing [ 14 C]benzoic acid at the specific activity of 1 ACi/mL.pH i was calculated as described under Materials and Methods.Determinations were done in quadruplicate for each condition, and the experiment was repeated at least thrice.Significant differences are based on the Student's t test (**, P < 0.01; ***, P < 0.005).D, DIDS, an inhibitor of HCO 3 À transporters, abolishes the differential in CAIX-induced alkalinization.The steady-state pH i of PS120-pev and PS120-pca9 cells was measured with [ 14 C]benzoic acid.Exponentially growing cells were incubated for 15 min in a nominally HCO 3 À Àfree MBS at a pH o of 6.6 without (DMSO ) or with 1 mmol/L DIDS.Cells were then shifted for 15 min to the same medium containing [ 14 C]benzoic acid at a specific activity of 1 ACi/mL.pH i was calculated as described under Materials and Methods.The mean of quadruplicate determinations is given.Significant differences are based on the Student's t test (**, P < 0.01).
were dissociated in Accutase (Life Technologies) to determine the number of individualized living cells.

Nude Mice Tumorigenicity and Immunohistochemistry
Cells (1 Â 10 6 ) suspended in 500 AL of serum-free DMEM supplemented with insulin-transferrin-selenium (Life Technologies) were s.c.injected into the back of 4-wk-old male athymic mice (Harlan).Animal studies were conducted according to Centre National de la Recherche Scientifique institutional guidelines.Food and water were given ad libitum.Doxycycline (Dox; 750 Ag/mL; Sigma) was given in the drinking water to induce silencing of hif-1a or ca9.Five mice were used for each experimental condition.The tumor volume was determined using the formula: (4p / 3) Â L / 2 Â W / 2 Â H / 2 (L, length; W, width; H, height).When the tumor volume reached f1,500 mm 3 , mice were injected i.p. with hypoxyprobe (60 mg/kg; Chemicon) 4 h before sacrifice.Tumors were collected for RNA, protein, and immunohistochemical analysis, as described (34).Sections were incubated with antibodies to hypoxyprobe or CAXII for 1.5 h followed by incubation with anti-mouse or anti-rabbit IgG-HRP antibodies.Analysis was performed with a Leica microscope (objective, 20Â).

Statistical Analysis
The Student's t test was used wherein P values of <0.05 were considered significant.

Results
Forced expression of catalytically active CAIX-induced extracellular acidification and cytoplasmic alkalinization.Nonneoplastic Chinese hamster CCL39 lung fibroblasts and CCL39-derived PS120 mutant cells defective in the Na + / H + exchanger, which do not express endogenous CAIX or CAXII in normoxia or in hypoxia, were selected for expression of human CAIX and examined for their contribution to pH i with (CCL39 cells) or without (PS120 cells) interference from pH i regulation by NHE-1.Stable expression of human CAIX in normoxia was found to be plasma membrane located in PS120 (Fig. 1A) and CCL39 cells (Supplementary Fig. S1).The level of expression of transfected human CAIX in these fibroblasts in Figure 2. CAIX expression promotes survival and growth of PS120 fibroblasts impaired in NHE-1 in vitro in an acidic microenvironment.A, proliferation of PS120-pev and PS120-pca9 cells (two independent clones for each type) was measured after a 24-h incubation in normoxia in a CO 2 /HCO 3 À Àfree microenvironment for media buffered with either 30 mmol/L MES for a pH o of 6.2, 6.4, 6.6, and 6.8 or with 30 mmol/L HEPES for a pH o of 7.4 and supplemented with 100 Amol/L hypoxanthine/100 Amol/L uridine/10% dialyzed FCS.A cell proliferation ELISA colorimetric kit, which measures BrdUrd incorporation into newly synthesized DNA, was used.Data are expressed as an average of three independent experiments performed in triplicate using two independent clones.Significant differences are based on Student's t test (***, P < 0.005).B, quantification of ATP levels in PS120-pev and PS120-pca9 cells (two independent clones for each type) grown in normoxia in a CO 2 /HCO 3 À Àfree environment with varying pH o from 6.2 to 7.4 for 24 h using a luciferin/luciferase-based assay, with results expressed as relative luminescence units.Determinations were done in triplicate, and the entire experiment was done thrice using two independent clones.Significant differences are based on the Student's t test (***, P < 0.005).C, viability assay of PS120-pev and PS120-pca9 cells in normoxia as function of pH o .Top, cells (1,000) were seeded in 60-mm dishes.Once attached, cells were incubated in HCO 3 À Àfree media adjusted to pH o 6.4 or 7.4 for 24 h in a CO 2 -free atmosphere.Fresh medium containing HCO 3 À (40 mmol/L) at pH o 7.4 was then added to the dishes that were maintained for 10 d in a 5% CO 2 incubator before staining for visualization of the colonies.Bottom, the colony number per plate was counted by eye for two independent PS120-pca9 and PS120-pev clones.The experiments were repeated thrice using two independent clones.Student's t test (*, P < 0.05).Bottom, total extracts were analyzed by immunoblotting with antibodies against HIF-1a, CAIX, CAXII, and Hsp90.The latter was used as a loading control.B, top, rate of acidification of a medium set at pH 8.2 in response to the addition of CO 2 incubated without cells (no cells ) or with intact LS-shca9/ctl cells previously incubated for 4 d with (ÀTet ) or without (+Tet ) Tet and exposed in hypoxia 1% O 2 for 48 h.Control LS-shca9/ctl ÀTet and ca9 silenced LS-shca9/ctl +Tet cells were also treated with 100 Amol/L ACTZ 15 min before addition of CO 2 .Bottom, rate of acidification in response to a CO 2 load of intact LS-shca9/ctl and LS-shca9/ca12 À cells incubated without (ÀTet) or with (+Tet) Tet for 4 d and then exposed to either normoxia (N ) or hypoxia 1% O 2 (H 1%) for 48 h.C, combined ca9 and ca12 invalidation (LS-shca9/ca12 À +Tet) diminishes resting pH i in an acidic (pH o 6.0), MES-buffered, nominally HCO 3 À Àfree solution.Stable expressing LS-shca9/ca12 À cells were constitutively invalidated for ca12, whereas exponentially growing Tet-inducible LS174Tr (LS-shca9/ctl and LS-shca9/ca12 À ) clones were invalidated for ca9 after a 4-d treatment with Tet.Cells were then incubated in hypoxic incubation 1% O 2 for 48 h.Dishes were then returned to normoxia for resting pH i determination in a 20 mmol/L MES-buffered, nominally HCO 3 À Àfree saline/glucose solution adjusted to pH o 6.0 or in a 20 mmol/L HEPES-buffered, nominally HCO 3 À Àfree saline/glucose solution adjusted to pH o 7.4 (15 min incubation to reach the equilibrium).For pH i determination, cells were shifted for 15 min to the same medium containing [ 14 C]benzoic acid at a specific activity 1 ACi/mL.The pH i was calculated as described under Materials and Methods.The experiment was done twice.Each point represents the average of quadruplicates for each experiment (**, P < 0.01).D, silencing of ca9 and ca12 diminished proliferation of three-dimensional spheroids.Top, an inducible LS174Tr clone invalidated for ca12 (LS-shca9/ca12 À ÀTet) or both ca9 and ca12 (LS-shca9/ca12 À +Tet) with a 4-d Tet treatment (+Tet ) were cultured as spheroids in the absence (ÀTet) or presence (+Tet ) of Tet for 12 d.Hypoxyprobe (pimonidazole) was added to the extracellular media for staining of hypoxic zones 4 h before formaldehyde fixation.Bottom, spheroids were subjected to Accutase dissociation, and individualized live cells were counted using Trypan blue.The spheroid proliferation index was calculated as the ratio of the cell number counted at day 12 to the cell number at day 0. Data represent the average of three independent experiments.normoxia was comparable with the expression of CAIX in tumor cells after 48 h in hypoxia 1% O 2 (data not shown).The CA activity associated with the plasma membrane was determined by the rapid acidification of a minimally buffered medium in response to addition of a CO 2 -saturated solution.In the presence of cell suspensions, the rate and magnitude of acidification was higher for CAIX-expressing PS120 cells (Fig. 1A) and CAIXexpressing CCL39 cells (Supplementary Fig. S1) than for control cells (pev).The CA inhibitor ACTZ reduced the activity of CAIXexpressing cells to the spontaneous basal level obtained with the control cells with or without ACTZ.In addition, PS120-pca9 cells incubated in normoxia or hypoxia (48 h, 1% O 2 ) showed the same level of activity (data not shown).The resting pH i of PS120-pca9 cells, with the BCECF-AM dye, showed it to be more alkaline compared with control PS120-pev cells when incubated in a nominally HCO 3 À /CO 2 -free solution set at a pH o of 6.6 to 7.4 (Fig. 1B).The difference was more pronounced for a pH o of 6.6 (0.4 units) compared with 7.4 (0.15 units).In the presence of 25 mmol/L bicarbonate, no difference in pH i was observed between CAIX-expressing cells and control cells.The intracellular alkalinization associated with CAIX expression was confirmed in PS120 (Fig. 1C) and CCL39 (Supplementary Fig. S2) cells with another technique that uses [ 14 C]benzoic acid, indicating that CAIX expression protects cells against cytoplasmic acidification.In the presence of endogenous NHE-1 (CCL39 cells), CAIX was able to restore a more alkaline pH i in acidic environments.This was more marked in fibroblasts impaired in NHE-1 because CAIX was able to compensate for the lack of NHE-1 in maintaining pH i in acidic environments.The implication of bicarbonate transport in changes in pH i is shown by its suppression in the presence of the bicarbonate transport inhibitor DIDS (Fig. 1D).This shows that the function of CAIX as a pH i regulator is revealed only in the absence of added extracellular bicarbonate.
CAIX-mediated cytoplasmic alkalinization and cell survival in an acidic environment.To examine if hypoxia-induced CAs protect cells from extracellular acidosis, we assessed the effect of forced expression of CAIX in PS120 and CCL39 cells on proliferation, ATP level, and cell survival.Cell proliferation determined by BrdUrd incorporation into DNA was significantly increased in normoxic CAIX-expressing cells at low pH o (Fig. 2A and Supplementary Fig. S3), as was the production of ATP (Fig. 2B and Supplementary Fig. S4) after 24 h in the absence of CO 2 /HCO 3 À .
Cell colony formation was not significantly different when comparing CAIX-expressing and control cells incubated at a pH o of 7.4; however, it was substantially diminished in non-CAIXexpressing cells at a low pH o of 6.4 (Fig. 2C and Supplementary Fig. S5).Stable PS120 clones expressing human CAXII were also obtained and showed similar characteristics to those expressing CAIX: extracellular acidification, conserved pH i regulation, and increase in cell survival in acidic conditions (data not shown).However, the CAXII activity was slightly lower compared with that of CAIX.These results suggest that CAIX and CAXII, by maintaining a more alkaline resting pH i , sustain ATP levels, promoting cell survival in a bicarbonate-free acidic microenvironment.Hypoxia-induced CAIX and CAXII activity contributes to cytoplasmic alkalinization in an acidic microenvironment.A number of human tumor cell lines showed an increase in the number of copies of ca9 and ca12 mRNA and CAIX and CAXII protein expression in response to hypoxia 1% O 2 , including RCC4, HeLa, A375Tr, A549, and LS174Tr cells (Supplementary Table S1; Supplementary Fig. S6).To further investigate the contribution of both CAIX and CAXII, we chose to use the Tet-inducible LS174Tr human colorectal adenocarcinoma cells to silence hif-1a with shRNA.In LS-shhif-1a cells, comparable levels of expression of hif-1a mRNA were detected in normoxia and hypoxia, confirming absence of transcriptional regulation of hif-1a by oxygen.In contrast, incubation of hypoxic LS-shhif-1a cells with Tet resulted in a substantial decrease in the mRNA level of hif-1a (Fig. 3A, top).Expression of ca9 and ca12 was significantly decreased in hypoxic LS-shhif-1a cells when hif-1a was silenced (+Tet; Fig. 3A, bottom).The LS174Tr cells endogenously express both membrane-bound and catalytically active CAs in a HIF-1-dependent manner (Fig. 3B).Hypoxic induction was associated with significantly enhanced alkalinization of the resting pH i of cells when exposed to a low pH o in the absence of extracellular bicarbonate (Fig. 3C).To study the contribution of CAIX to pH i regulation, hypoxia-induced CAIX expression was mimicked in LS174Tr cells expressing a basal level of endogenous CAXII in normoxia (Fig. 3D; Supplementary Table S1; Supplementary Fig. S7).Forced Tet-inducible expression of CAIX leads to enhanced alkalinization of the resting pH i in an acidic and bicarbonate-free environment (Fig. 3D and Supplementary Fig. S7).Thus, CAIX expression plays a key role in maintaining the resting pH i in an acidic and bicarbonate-limiting environment in LS174Tr cells.
Combined invalidation of CAIX and CAXII reduces pH i and spheroid growth.LS174Tr cells were selected for Tet-inducible silencing of ca9 combined (LS-shca9/ca12 À ) or not (LS-shca9/ctl) with constitutive silencing of ca12.Tet addition to shca9 cells (LS-shca9/ctl and LS-shca9/ca12 À ) resulted in 95% invalidation of ca9 mRNA (Fig. 4A, top) and protein (Fig. 4A, bottom), whereas 90% of the ca12 mRNA (Fig. 4A, top) and protein (Fig. 4A, bottom) were silenced in LS-shca9/ca12 À cells.Note that an increase in hypoxiainducible expression of the mRNA and protein levels of CAXII was observed when ca9 was silenced.The copy number of ca9 mRNA is 18-fold lower than that of ca12 in LS-shca9/ctl in normoxia (Supplementary Table S2), whereas in hypoxia, the number of copies of ca9 mRNA was only twice higher than that of ca12.Moreover, when ca9 is suppressed in hypoxia, the number of ca12 mRNA copies was almost similar to ca9 induced in hypoxia.In addition, when ca9 was silenced in hypoxia, the overall hypoxia-induced CA activity was unchanged whereas ACTZ showed a marked reduction in acidification (Fig. 4B, top).When ca12 was silenced in hypoxia, no reduction in extracellular acidification occurred (data not shown); however, combined silencing of ca9 and ca12 in hypoxia (LS-shca9/ca12 À +Tet) reduced the CA activity to the basal level (Fig. 4B, bottom).
We then determined the relative contributions of CAIX and CAXII in hypoxia regarding pH i regulation.At a neutral pH o of 7.4, silencing of ca9, ca12, or both together did not affect the resting pH i (Fig. 4C).When cells were exposed to an acidic pH o of 6.0, only combined invalidation of ca9 and ca12 resulted in a significantly lower resting pH i (0.2 units), whereas silencing of either isoform alone had no effect (Fig. 4C); results were obtained with two independent sequences targeting ca12 (data not shown for the second sequence).We then assessed the importance of this pH i regulating system in hypoxia on cells grown as three-dimensional spheroids.In a bicarbonate-limiting environment, a hypoxic gradient is established (hypoxyprobe labeling; Fig. 4D, top) and, thus, lactic acid is produced.When ca9 was silenced (LS-shca9/ctl +Tet) the proliferation index of hypoxic spheroids diminished compared with control cells (LS-shca9/ctl ÀTet) and diminished further when both isoforms were silenced (LS-shca9/ca12 À +Tet; Fig. 4D, bottom).These results indicate that both CAIX and CAXII play an important role in the regulation of pH i recapitulating the protective effect of hypoxia in promoting cell survival in an acidic environment.
Inducible invalidation of CAIX and CAXII reduces the rate of xenograft tumor growth.To investigate the in vivo functional consequence of CAIX and CAXII expression on tumor growth, athymic mice were s.c.injected with Tet-inducible LS-shhif-1a, LS-shca9/ctl, or LS-shca9/ca12 À cells.Dox had no significant effect on tumor growth of control cells (LS-shev/ctl; Fig. 5A, top), whereas Dox silencing of hif-1a showed a substantial decrease in the size of tumors (Fig. 5A, bottom).The constitutive silencing of ca12 in an Figure 6.Immunohistologic confirmation of the expression of HIF-1a, CAIX, or CAXII in hypoxic regions of the corresponding tumor xenografts.Mice were injected with ca9-deficient LS-shca9/ctl cells (A ) or ca9 -and ca12 -deficient LS-shca9/ca12 À cells (B) and drinking water supplemented without (ÀDox ) or with (+Dox ) Dox was provided to mice.Serial sections of tumors of the same size were stained for hypoxic regions with antibodies to hypoxyprobe (pimonidazole), for nuclear staining of HIF-1a, or for membrane staining of CAIX or CAXII.Costaining of tumor sections was done first using the immunofluorescence protocol for CAIX followed by immunohistochemistry for CAXII in the same tumor section of LS-shca9/ctl cells (C ) and LS-shca9/ca12 À (D ).Magnification, 20Â.V, major blood vessel of the tumor lobe analyzed in serial sections.

Cancer Research
Cancer Res 2009; 69: (1).January 1, 2009 endogenous ca9 background (LS-shca9/ca12 À ÀDox) showed similar tumor growth to that of the control cells (LS-shca9/ctl ÀDox; Fig. 5B).However, the silencing of ca9, in the presence of endogenous ca12 (LS-shca9/ctl +Dox), gave a slight but significant reduction in tumor growth (40%) compared with control cells (Fig. 5C).Invalidation of both isoforms (LS-shca9/ca12 À +Dox) resulted in a spectacular decrease in tumor size (85%) as a result of slower cell proliferation (Fig. 5C).Examination of the mRNA levels of ca9 and ca12 in the tumors confirmed almost complete invalidation of ca9 in the invalidated cell lines when Dox was added (95% for LSshhif-1a, 94% for LS-shca9/ctl, and 92% for LS-shca9/ca12 À ).Whereas the expression of the ca12 mRNA was diminished by a half in hif-1a invalidated cells (Figs. 4A and 5D), its expression increased 1.5-fold when ca9 (LS-shca9/ctl +Dox) was silenced (Fig. 5D).In the LS-shca9/ca12 À cell line, the level of ca12 mRNA was diminished (80%) whereas that of ca9 in the absence of Dox was not significantly different.These results indicate that the respective tumor types retain their invalidated phenotype and that the expression of ca12 responds to the level of expression of ca9, although the reverse is not the case.
Immunohistochemical and immunofluorescence analysis of tumor sections from mice injected with LS-ca9/ctl (Fig. 6A and C) and LS-ca9/ca12 À cells (Fig. 6B and D) showed diminished levels of CAIX in cells silenced for ca9 and for ca9 plus ca12 in Dox-treated mice, despite the expression of HIF-1a in hypoxic zones (hypoxyprobe staining).CAXII expression is seen in all cells of the tumor (slightly increased in the perinecrotic area) in contrast to the expression of CAIX, which correlates with the hypoxic and perinecrotic regions.It is important to note the increase in CAXII expression when CAIX is silenced in vivo.Comparison of the immunohistochemistry of tumor sections of control LS-shev/ctl cells (Supplementary Fig. S8A) and hif-1a silenced LS-shhif-1a cells (Supplementary Fig. S8B) revealed a significant decrease in HIF-1a, CAIX, and CAXII in hif-1a silenced cells.For tumor histology, to insure that the reduced labeling observed for CAIX in ca9/ca12 À silenced tumors was indeed due to invalidation and not small-sized tumors with minimal hypoxic zones, mice were maintained for a longer time and sacrificed 27 days later than control mice when the tumors were the same size as for control nonsilenced cells.Thus, these results show that invalidation of both CAIX and CAXII brings about a dramatic decrease in tumor xenograft cell growth.

Discussion
Investigation into the implication of hypoxia-inducible CAs in the regulation of pH has concerned only CAIX and has been restricted to monolayer or three-dimensional cell cultures (31,32).During the preparation of this manuscript, a study showed that ectopically expressed CAIX in human bladder carcinoma RT112 cells was able to spatially coordinate pH i , but only when cells are cultured as three-dimensional spheroids (32).In agreement with this report, CAIX expression in our study had no effect on pH i regulation in isolated cells in a neutral and bicarbonate-buffered medium (25 mmol/L); however, when cells were exposed to a nominally bicarbonate-free and acidic milieu, CAIX effected on the resting pH i .Here, we conducted in vitro studies in the absence of extracellular bicarbonate in order not to saturate bicarbonate transporters at the cell surface.As we showed previously (6,35), the presence of a high bicarbonate level (25 mmol/L) totally blunts the effect of NHE-1 on pH i regulation.Thus, we reasoned that, to investigate the putative contribution of membrane-bound CAIX and CAXII in pH i regulation, it is necessary to operate in nominally bicarbonate-free solutions exposed only to ambient CO 2 .
We also examined CAIX regulation of pH i in fibroblasts impaired in NHE-1 expression (PS120 cells) to ensure the absence of interference by this major player in pH i homeostasis (12,(36)(37)(38).Nonetheless, the effect of CAIX on pH i was also detected in CCL39 fibroblasts and in a human colon adenocarcinoma cell line LS174Tr expressing endogenous NHE-1.Previous studies showed that NHE-1-deficient cells fail to grow in a range of acidic pH o (6.2-6.8)due to their inability to reach the permissive pH i values required for DNA synthesis and ATP production (6,7,39).Forced CAIX expression in this pH i regulation-deficient cell system was able to restore viability of PS120 cells when exposed to a range of acidic pH o (6.2-6.8).This shows the role of a pH i -threshold value for growth confirming the role of CAIX in pH i control.
It is postulated that the mechanism by which membrane-bound CAs regulate pH i occurs through the efficient uptake of HCO 3 À locally formed in the ''mouth'' of CAs through Cl À /HCO 3 À exchangers and/or Na + /HCO 3 À cotransporters, forming tight functional complexes (40).Although the ''metabolon'' is an interesting concept, this notion has been challenged (41)(42)(43).Interaction of the catalytic domain of CAIX with bicarbonate transporters has also been reported and was shown to increase the AE exchanger activity (21,44).Future investigation is under way to evaluate the key bicarbonate transporters coupled to CAIX and CAXII in pH i regulation in hypoxic tumor cells.
In our study, invalidation of CAIX leads to partial compensation by up-regulation of CAXII.This may explain the maintenance of the catalytic activity of ca9 invalidated cells in hypoxia and suggest that a threshold level of activity is required for cell pH homeostasis.In the recent study of Swietach and colleagues (32), overexpression of CAIX down-regulated cytosolic CAII.This result suggests that different tissues with different expression patterns of CAs may bring into play different CA isoforms that would ''communicate'' in a yet unresolved network when confronted with an acidic stress.
Hypoxic induction may not be the only mechanism by which CAs regulate pH homeostasis in tumors.Signaling through the epidermal growth factor pathway by phosphorylation of a cytoplasmic tyrosine residue of CAIX may either activate CAIX or enhance its expression by increasing translation of HIF-1a (30).In addition, phosphorylation activates phosphatidylinositol 3-kinase, resulting in phosphorylation of Akt and cell survival.The possible implication of these CAs as signaling molecules, independent of their function as pH-regulating enzymes, is another important point that is under investigation and may explain reduced xenograft growth of ca9 silenced cells despite maintenance of CA catalytic activity.
It has been proposed that an acidic microenvironment promotes metastasis associated with poor patient survival (4,45).However, acidosis may not always favor metastasis (4,46).When renal carcinoma cells were treated with ACTZ, their capacity to invade was diminished (47) but the invasion of carcinoma cells was not influenced by CAIX in another study (48).Here, we not only show that CAIX and CAXII promote survival in an acidic environment in two cell culture systems but extend it in in vivo studies.The combined silencing of CAIX and CAXII gave a dramatic decrease in the rate of growth of xenograft tumors that was greater than the invalidation of HIF-1a, which may reflect the pleotropic action of HIF-1 on prosurvival and prodeath genes.This may suggest that the interest being paid to HIF inhibitors as anticancer treatments (49,50) might be better directed to the inhibition of downstream HIF target gene products, as we recently proposed (1) and, in particular to CAIX together with CAXII, as novel and potentially efficient drug approaches.
The present study highlights the role of CAIX and CAXII expression in pH i regulation, a key event controlling cell viability, and in in vivo tumor growth in a hostile acidic and hypoxic microenvironment.The results herein call for the development of specific ''CA-antagonist'' antibodies or cell impermeable drugs specifically targeting the membrane-associated and hypoxiainducible CAs, taking into consideration the CA isoform profile of a given tumor type.Such inhibitors are being actively investigated at the molecular and cellular levels (23,24) and may hold promise as effective anticancer treatments.

Figure 1 .
Figure 1.Forced expression of catalytically active CAIX induced extracellular acidification and cytoplasmic alkalinization.A, top, stable expression of human CAIX in nonneoplastic PS120 fibroblasts lacking the Na + /H + exchanger (NHE-1).Inset 1, immunoblot to CAIX of lysates of control, empty plasmid vector (PS120-pev ), or human CAIX stable expressing cells (PS120-pca9 ).The Hsp90 protein level was used as a loading control.Inset 2, immunofluorescence to CAIX in PS120-pev and PS120-pca9 cells.Bottom, rate of extracellular acidification in response to a CO 2 load without cells (no cells ) or with samples of intact normoxic PS120-pev or PS120-pca9 cells.The rate of acidification reflects an estimation of the CA activity at the cell surface evaluated with or without a specific inhibitor of CAs (100 Amol/L ACTZ).B, CAIX expression increases the resting pH i .pH i of PS120-pev and PS120-pca9 cells was determined with pH-sensitive fluorescent dye BCECF-AM.Histograms indicate the mean basal pH i for cells incubated for 30 min in either a 25 mmol/L HCO 3 À -buffered solution (+HCO 3 À ) or HCO 3 À -free MES or HBS (-HCO 3 À

Figure 3 . 3 À
Figure 3.The hypoxia-induced CAIX and CAXII activity contributes to cytoplasmic alkalinization in an acidic microenvironment.A, the genes ca9 and ca12 are HIF-1-dependent in human colon adenocarcinoma LS174Tr cells.Expression of the mRNA of hif-1a (top ), ca9 , and ca12 (bottom ) was determined by real-time qPCR in the stable LS-shhif-1 a clone, in which Tet induces shRNA of hif-1a .Cells were incubated without (ÀTet ) or with (+Tet ) Tet for 4 d and maintained in either normoxia 21% O 2 (N ) or hypoxia 1% O 2 (H 1%) for 48 h.Each difference in gene expression was estimated within the limits of a 95% confidence interval.The results are representative of at least three separate experiments.B, hypoxia increases the extracellular CA activity, as reflected in an increase in the rate of extracellular acidification compared with normoxic conditions.Top, LS174Tr cells were incubated either in normoxia or hypoxia 1% O 2 for 48 h.The rate of extracellular acidification in response to a CO 2 load was measured without cells (no cells ) or with intact LS174Tr cells.Bottom, The LS-shhif-1a clone was incubated without (ÀTet ) or with (+Tet ) Tet to allow for hif-1a silencing for 4 d before incubation in hypoxia 1% O 2 for 48 h without (ÀTet ) or with (+Tet ) Tet.The rate of acidification in response to a CO 2 load for intact hypoxic LS-shhif-1a cells silenced (+Tet ) or not (ÀTet) for hif-1a was measured.ACTZ (+ACTZ ) at 100 Amol/L was added to a suspension of hypoxic LS-shhif-1a ÀTet cells for 15 min before CO 2 addition.C, determination of resting pH i with [ 14 C]benzoic acid as a function of pH o in LS174Tr cells exposed to either normoxia or hypoxia 1% O 2 for 48 h.Exponentially growing LS174Tr cells incubated in normoxia or hypoxia were equilibrated for 15 min in a nominally HCO 3 À Àfree HBS or MBS with a pH o varying from 5.8 to 7.4.Cells were incubated for 15 min in the same appropriate equilibration solution containing [ 14 C]benzoic acid at a specific activity of 1 ACi/mL.The pH i was calculated as described under Materials and Methods.The experiment was repeated at least thrice, and each point represents the average of quadruplicates for each experiment (**, P < 0.01; ***, P < 0.005).D, stable Tet-inducible expression of CAIX in LS174Tr cells in normoxia maintains a higher pH i in an acidic and HCO 3 À Àfree media.Top, inset 1, immunoblot of inducible expression of CAIX in normoxia in LS174Tr cells (LS-pca9 ) without (ÀTet) or with (+Tet ) Tet for 4 d.Total extracts were analyzed by immunoblotting with antibodies against CAIX and Hsp90.The latter was used as a loading control.Inset 2, immunofluorescence of inducible expression of CAIX expression (LS-pca9 ) in LS174Tr cells in the absence (ÀTet) or presence (+Tet ) of Tet for 4 d.Bottom, determination of resting pH i with [ 14 C]benzoic acid in as a function of pH o in Tet-inductible LS-pca9 clone expressing CAIX in normoxia under treatment for 4 d with (+Tet ) or without (ÀTet ) Tet. Determinations were done in quadruplicate for each condition, and each experiment was repeated at least thrice.Results represent the average of quadruplicates for each experiment (**, P < 0.01; ***, P < 0.005).

Figure 4 .
Figure 4. Combined invalidation of CAIX and CAXII reduces pH i and spheroids growth.A, silencing of ca9, ca12 , and ca9/ca12 in LS174 cells.Top left, expression of the mRNA levels of ca9 determined by real-time qPCR.Stable LS-shca9/ctl and LS-shca9/ca12 À clones were incubated with (ÀTet ) or without (+Tet ) Tet for 4 d before incubation in either normoxia (N ) or hypoxia 1% O 2 (H 1%) for 48 h.Top right, expression of the mRNA levels of ca12 determined by real-time qPCR under the conditions described below.LS-shca9/ctl and LS-shca9/ca12 À clones were incubated with (ÀTet ) or without (+Tet ) Tet for 4 d before incubation in either normoxia (N ) or hypoxia 1% O 2 (H 1%) for 48 h.The results are representative of at least three separate experiments (**, P < 0.01).Bottom, total extracts were analyzed by immunoblotting with antibodies against HIF-1a, CAIX, CAXII, and Hsp90.The latter was used as a loading control.B, top, rate of acidification of a medium set at pH 8.2 in response to the addition of CO 2 incubated without cells (no cells ) or with intact LS-shca9/ctl cells previously incubated for 4 d with (ÀTet ) or without (+Tet ) Tet and exposed in hypoxia 1% O 2 for 48 h.Control LS-shca9/ctl ÀTet and ca9 silenced LS-shca9/ctl +Tet cells were also treated with 100 Amol/L ACTZ 15 min before addition of CO 2 .Bottom, rate of acidification in response to a CO 2 load of intact LS-shca9/ctl and LS-shca9/ca12 À cells incubated without (ÀTet) or with (+Tet) Tet for 4 d and then exposed to either normoxia (N ) or hypoxia 1% O 2 (H 1%) for 48 h.C, combined ca9 and ca12 invalidation (LS-shca9/ca12 À +Tet) diminishes resting pH i in an acidic (pH o 6.0), MES-buffered, nominally HCO 3