Enhanced activation of an amino-terminally truncated isoform of the voltage-gated proton channel HVCN1 enriched in malignant B cells

Identification of an HVCN1 Isoform Enriched in Malignant B Cells.

We investigated HVCN1 expression in a panel of B-cell lines and noted the presence of two bands of similar molecular weight. Some cell lines appeared to express more of the shorter isoform, whereas the Burkitt cell line Raji expressed it exclusively (Fig. 1A, Left). Peripheral blood B cells expressed this short isoform (here called HVCN1_S) at much lower levels compared with CLL cells (Fig. 1A, Center and Right).

When we started this study, the GenBank database reported only one viable splicing variant for human HVCN1 (SI Appendix, Fig. S1, variant 1). Given the presence of an ATG 60 base pairs (bp) downstream from the first ATG, we set out to investigate if a shorter isoform could be the result of translation from this alternative start site. To this end, we expressed recombinant HVCN1 Long (HVCN1_L) in an in vitro translation assay (Fig. 1B, Left). We then mutated the first ATG to TTG, which resulted in significantly reduced translation from this codon (Fig. 1B, Center). A shorter protein was expressed from the mutated plasmid (Fig. 1B, Center) when expressed in whole cells also (Fig. 1C, Right), indicating that the second ATG functioned as a start codon.

More recently, three splicing variants for human HVCN1 have been reported (SI Appendix, Fig. S1). Variant 1 and 2 differ in the use of alternative 5′-untranslated regions (UTRs) but code for the same full-length HVCN1 protein, whereas variant 3 uses the same 5′-UTR as variant 1 but it does not possess the first coding exon. Therefore, variant 3 lacks the first ATG and consequently translation can start only from the second ATG 60 bp downstream. To test the existence of the three mRNA variants, we designed specific PCR primers and tested them with mRNA from the cell line HBL1. We used the same reverse primer and specific forward primers for variants 1/3 and 2, annealing on their respective 5′-UTRs. To distinguish variants 1 and 3, we relied on differences in the band size of the amplified PCR product (433 for variant 1 and 393 for variant 3). As Fig. 1C, Left, shows, the PCR with primers for variant 2 and all variants produced bands of the expected sizes. For variants 1 and 3, we ran a gradient PCR (Fig. 1C, Right), which showed the expected bands of 433 and 393 bp (Fig. 1C, arrows). Overall, our results indicate the existence of three distinct mRNA variants and therefore suggest that HVCN1_S is the result of alternative mRNA splicing, which produces a protein identical to the long isoform but lacking the first 20 amino acids.

To confirm the nature of this shorter isoform, we raised an antibody toward the first 20 amino acids of HVCN1, which should recognize the long isoform only. Our original anti-HVCN1 antibody, which recognizes a region in the N terminus closer to the first transmembrane domain (amino acids 26–46), detected both isoforms in CLL cells (Fig. 1D, Left). Also, recombinant HVCN1_L and HVCN1_S were both detected. In contrast, an immunoblot with the anti-HVCN1_L–specific antibody detected only the long isoform, both recombinant and endogenously expressed (Fig. 1D, Right). These results confirmed that the shorter isoform of HVCN1 is a protein with a shorter N terminus domain, lacking the first 20 amino acids.

Both Isoforms Are Voltage-Gated Proton Channels.

Biophysical properties of HVCN1_L and HVCN1_S were compared in transduced LK-35.2 cells using the whole-cell voltage-clamp configuration over a wide range of pH. The H⁺ currents generated by the two isoforms appeared generally similar. SI Appendix, Fig. S2A, confirms that both isoforms were proton selective because the reversal potential for current (V_rev) was close to the Nernst potential for H⁺ (E_H). This result is expected because the selectivity filter is in the middle of the S1 transmembrane helix (1), far from the N terminus where the two isoforms differ.

A unique property of all known voltage-gated proton channels crucial to their function is ΔpH-dependent gating (2, 7). The position of the proton conductance–voltage (g_H–V) relationship depends strongly on both pH_o and pH_i, with the consequence that the channel opens only when there is an outward electrochemical gradient, and the open channel will extrude H⁺ from the cell. HVCN1_L and HVCN1_S exhibited similar ΔpH-dependent gating, both activating at potentials about 10 mV positive to V_rev at symmetrical pH (SI Appendix, Fig. S2B).

To compare gating kinetics, we used the perforated-patch voltage-clamp method, a more physiological configuration that preserves cytoplasmic contents. Comparison of the gating kinetics of HVCN1_L and HVCN1_S in unstimulated LK-35.2 cells (SI Appendix, Table S1) reveals that HVCN1_S channels opened more slowly, with a time constant of current turn-on, τ_act, more than double that of HVCN1_L. Closing kinetics (τ_tail) and the position of the g_H-V relationship did not differ significantly.

HVCN1_S Responds More Strongly to PKC-Dependent Phosphorylation.

Proton currents in phagocytes and other cells are greatly augmented by phosphorylation of the channel by PKC (8). The enhanced gating response is stimulated effectively by the PKC activator PMA (phorbol myristate acetate) and is best studied using the perforated-patch configuration that preserves intracellular signaling pathways (9). Fig. 2 illustrates families of proton currents in cells expressing HVCN1_L and HVCN1_S before and after PMA stimulation. In response to PMA, the currents turn on more rapidly and at more negative voltages and turn off more slowly, and the current amplitude is increased. Although HVCN1_L responds distinctly, the response of HVCN1_S was consistently more profound. Because there is a tendency early in each experiment for proton currents to become larger and activate at more negative voltages as the amphotericin in the pipette solution improves electrical access to the cell membrane, and as pH_i is clamped to 7.0 by the applied NH₄⁺ gradient (9, 10), the PMA response may be exaggerated if measurements are made before complete equilibration. A crucial quantitative control is to reverse the effects of PMA using the PKC inhibitor GF 109203X (GFX). The reversal of enhanced gating by GFX in both representative cells in Fig. 2 was complete, validating the responses.

SI Appendix, Fig. S3, shows normalized g_H-V relationships after the PMA response and after GFX treatment in all cells studied expressing HVCN1_L or HVCN1_S. This comparison is more informative than control vs. PMA for reasons just discussed and because some cells were spontaneously active, as judged by GFX reversal being greater than the initial response to PMA. A possible spurious explanation for the greater PMA responsiveness of HVCN1_S is that HVCN1_L might have a greater tendency to activate spontaneously. SI Appendix, Fig. S3, does not support this view, verifying that HVCN1_S is indeed more responsive.

SI Appendix, Fig. S4, summarizes the magnitude of the changes in four kinetic parameters of HVCN1 gating resulting from PMA and GFX treatment of cells expressing the two isoforms. For each parameter, the PMA response was significantly greater in HVCN1_S than in HVCN1_L. Also for each parameter, GFX reversed the PMA response completely. In summary, HVCN1_S channels exhibit a more profound response to phosphorylation than do HVCN1_L channels.

Because B cells from CLL patients have higher-than-normal levels of HVCN1_S, their responsiveness might be enhanced. B cells from CLL patients and normal controls were studied by perforated-patch voltage clamp (n = 4 and 3, respectively), confirming that both CLL and normal B cells had proton currents that responded to PMA and GFX. Interestingly, CLL cells, which have a variable mixture of HVCN1_S and HVCN1_L (SI Appendix, Table S3), exhibited enhanced gating responses comparable with those of LK35.2 cells expressing HVCN1_S exclusively (SI Appendix, Fig. S5), indicating that HVCN1_S prevails in the response.

Identification of HVCN1_S Phosphorylation Site.

Two predicted high-probability PKC-δ phosphorylation sites exist in the N terminus of HVCN1_L (11), Thr²⁹ and Ser⁹⁷. We reported previously that both sites were phosphorylated after PMA exposure, but only Thr²⁹ contributed detectably to the enhanced gating response of HVCN1_L channels (11). Because the lack of the first 20 amino acids of the N terminus might result in different protein folding or architecture, it was important to examine the contribution of both corresponding sites in HVCN1_S, namely Thr⁹ and Ser⁷⁷. We generated mutants that lacked one (T9A and S77A) or both (T9A/S77A) putative phosphorylation sites.

Cells with the T9A mutation (n = 9) and the double mutant T9A/S77A (n = 5) did not respond detectably to either PMA or GFX (SI Appendix, Fig. S6), implicating Thr⁹ as crucial to the responses. In contrast, cells expressing the S77A mutant did respond to PMA and to GFX. Mean changes in parameter values are given in SI Appendix, Table S2. Therefore, the key residue in the PMA response is Thr⁹ (HVCN1_S), the equivalent of Thr²⁹ in HVCN1_L channels (11). Any participation of Ser⁷⁷ (HVCN1_S) or Ser⁹⁷ (HVCN1_L) in the enhanced gating response was below our ability to detect it.

Intriguingly, T9A cells studied before stimulation exhibited ninefold faster activation and a 24-mV more negative V_threshold than did unstimulated HVCN1_S cells (SI Appendix, Table S1), resembling enhanced gating mode behavior. We reanalyzed data from analogous mutations studied previously in the full-length channel (11) and found that T29A also exhibited “enhanced gating” in unstimulated cells (SI Appendix, Table S1). Unstimulated “phosphomimetic” T29D cells exhibited weaker enhanced gating than T29A. Evidently, this Thr position is a potent determinant of gating kinetics. Together, these results suggest that phosphorylation of Thr²⁹ in HVCN1_L or Thr⁹ in HVCN1_S produces enhanced gating by a mechanism not simply involving the negative charge provided by the phosphate group. However, we cannot rule out the possibility that these mutations alter gating through a mechanism unrelated to that elicited by phosphorylation.

Because HVCN1_S responded more strongly to PKC phosphorylation and the effect was mediated by phosphorylation of Thr⁹ exclusively, we speculated that it could reflect increased phosphorylation of this residue. Indeed, our in vitro kinase assay showed that overall phosphorylation of wild-type HVCN1_S was significantly higher than for wild-type HVCN1_L (Fig. 3, Left, WT lanes). To assess phosphorylation of the critical PKC site mediating the enhanced-gating response, we carried out the same experiment with T29A (T9A for HVCN1_S), S97A (S77A for HVCN1_S), and T29A/S97A (T9A/S77A for HVCN1_S) mutants. As shown in Fig. 3, phosphorylation of the T9A mutant was reduced as much as for the T29A one, confirming that this Thr is phosphorylated by PKC in both isoforms. The analysis of the other PKC site mutant, S97A (S77A for HVCN1_S), and double mutants indicated a further reduction in overall phosphorylation, as we reported previously for HVCN1_L (11), suggesting that Ser is also phosphorylated by PKC. However, as before, the patch-clamp experiments (SI Appendix, Fig. S6 and Table S1) indicated that phosphorylation of Thr⁹ alone is responsible for the enhancement of HVCN1_S proton currents.

HVCN1_S Has a Weaker Association with the BCR than Does HVCN1_L.

We found previously that HVCN1 was associated with the BCR in both normal and malignant B cells (3). The association causes HVCN1 to be internalized together with the BCR after its stimulation, an event that downregulates HVCN1 expression at the cell surface (3). Because the intracellular domains are likely to be the regions mediating the association with the BCR, we investigated if HVCN1_S behaved differently from HVCN1_L. We overexpressed both isoforms in the B lymphoma cell line A20 D1.3 and assessed their expression at steady state and after BCR stimulation with a F(ab′)₂ anti-IgM. Cells expressed similar levels of HVCN1, IgM, and the BCR coreceptor CD79B (SI Appendix, Fig. S7 A–C). We did not note significant differences in the pattern of expression of the two isoforms at steady state. Importantly, however, upon BCR stimulation, the extent of HVCN1 internalization for the two isoforms appeared different (Fig. 4A). Quantification of the immunofluorescence staining showed a significantly larger percentage of internalized HVCN1 in cells expressing HVCN1_L compared with HVCN1_S (36 vs. 25%, Fig. 4A, Upper graph). Furthermore, the extent of colocalization with IgM was also reduced from 0.562 to 0.407 (Fig. 4A, Lower graph). Diminished HVCN1_S internalization was not due to reduced IgM internalization, which was actually increased, compared with cells overexpressing HVCN1_L (SI Appendix, Fig. S7D). These data indicate that HVCN1_S association with the BCR is different and that this results in increased localization of proton channels at the plasma membrane.

To determine the reason for the different association of HVCN1 isoforms with the BCR, we carried out a coimmunoprecipitation (co-IP) with an anti-CD79B and the reciprocal experiment with an anti-myc (for the myc-tagged HVCN1 isoforms). As shown in Fig. 4B, interaction of HVCN1_S with CD79B was indeed significantly weaker with both co-IP assays. The differences between the two antibodies are likely due to differences in the efficiency of pull down (with the anti-myc being more efficient than the anti-CD79B). Overall, these data show that the first 20 amino acids present in HVCN1_L are critical for the association with the BCR and that their absence in HVCN1_S results in greater expression of this isoform at the plasma membrane after BCR stimulation.

HVCN1_S Expression Is Enriched in CLL Cells.

Given the increased expression of HVCN1_S in B-cell lines and in CLL samples (Fig. 1A), combined with its ability to mediate stronger proton currents (Fig. 2), we investigated whether its expression could have an impact on CLL disease progression. To this end, we analyzed the expression of HVCN1_S in a cohort of 76 samples of peripheral blood CLL cells with annotated clinical data. Details of the patients’ characteristics are reported in SI Appendix, Table S3. The samples investigated showed variable expression of both isoforms, with an average ratio of HVCN1_S to HVCN1_L of 0.52 ± 0.034 (mean ± SE). As shown in Fig. 1A, HVCN1_S was markedly higher in CLL than in primary B cells from healthy donors, in which HVCN1_S is barely detectable. Given the wide range of HVCN1_S expression in CLL cells, we split samples in two groups, one with a ratio of HVCN1_S/HVCN1_L below the median value of 0.477 (i.e., with lower expression of HVCN1_S) and one with higher HVCN1_S expression (above the median). A Kaplan–Meier curve for overall survival showed that patients with higher expression of HVCN1_S had reduced overall survival, although differences did not reach statistical significance (SI Appendix, Fig. S8). We further analyzed the data specifically for the two main subgroups of CLL patients: those presenting a mutated variable region in the BCR heavy chain (IGHV) and those with an unmutated IGHV. Given the small number of samples for which we knew the mutation status (SI Appendix, Fig. S8 and Table S3), it was not surprising that differences did not reach statistical significance; nonetheless, higher expression of HVCN1_S tended to correlate with a poorer outcome in both groups.

HVCN1_S Expression Results in Stronger BCR-Dependent Signaling, Proliferation, and Chemokine-Dependent Migration.

To establish if HVCN1_S could confer a growth advantage to malignant B cells, we first assessed BCR-dependent signaling in A20 D1.3 cells transduced with empty vector (EV), HVCN1_L, and HVCN1_S. As shown in Fig. 5A, the presence of HVCN1_S resulted in more prolonged activation of the extracellular signal-regulated kinase (Erk), which is responsible for the upregulation of several antiapoptotic proteins in CLL cells (12). Furthermore, to determine if the two proton channel isoforms regulated cell proliferation, we measured the extent of EdU incorporation, a nucleoside analog included in nascent DNA. Interestingly, only HVCN1_S provided a significant advantage compared with EV cells (Fig. 5B).

Another important property of CLL cells is the ability to respond to the chemokine CXCL12 [chemokine (C-X-C motif) ligand 12], which affects their migration, homing, and survival (13). Because HVCN1 was shown to regulate migration of breast (5) and colorectal cancer cells (6), we set out to characterize if HVCN1 isoforms affected chemokine-dependent migration in a B-cell setting. As Fig. 5C shows, both HVCN1_L and HVCN1_S expression resulted in increased migration to CXCL12; however, only HVCN1_S resulted in a significant advantage. Taken together, these data indicate that HVCN1_S promotes malignant B-cell survival through enhanced proliferation and migration.

Cell Lines and Plasmids.

The mouse B lymphoma cell lines LK35.2 HyHEL10 (IgG2a, κ-chain; H-2^kxd) and A20 D1.3 (IgG2a, κ-chain; H-2^d) overexpressing recombinant IgM receptors were a gift from F. Batista (London Research Institute, London). The cells were transduced with MigRI retroviral vectors coding for myc-tagged HVCN1_L or HVCN1_S. T29A (T9A for HVCN1_S), S97A (S77A for HVCN1_S), and T29A/S97A (T9A/S77A for HVCN1_S) mutants were generated using a QuikChange site-directed mutagenesis kit (Stratagene). The retroviral particles were produced in the Phoenix α packaging cell line as described elsewhere (11).

Phosphorylation Assay.

HEK293T cells were transfected by Ca²⁺ phosphate with myc-tagged HVCN1_L, HVCN1_S, and mutant plasmids. Forty-eight hours after transfection, cells were lysed in 20 mM Hepes, 1% Triton X-100, 137 mM NaCl, 2.5 mM β-glycerophosphate, 1 mM Na₃VO₄, 2 mM EDTA, and proteases inhibitors (Sigma Aldrich). One milligram of proteins was immunoprecipitated with anti-myc tag antibody (9B11, Cell Signaling Technology) conjugated to protein-G Sepharose beads for 1 h. After washing with lysis buffer, beads were incubated in 40 μL kinase assay buffer [20 mM Hepes, 1 mM EGTA, 0.4 mM EDTA, 5 mM MgCl₂, 0.1 mM CaCl₂, 0.05 mM DTT, 0.1 mg/mL phosphatidylserine, 0.01 mg/mL diacylglycerol, 2.5 mM β-glycerophosphate, 1 μM PMA, 100 nM PKC-δ (Millipore), 100 μM cold ATP, and 10 μCi of [γ-³²P]ATP (Perkin-Elmer)] for 20 min at 30 °C. The reaction was stopped by resuspending beads in 2× Laemmli sample buffer. Samples were then separated by SDS/PAGE, transferred to a nitrocellulose membrane, and exposed to X-ray films. Membranes were immunoblotted with anti-myc antibody to determine loading.

Electrophysiology.

Whole-cell or perforated-patch variants of the patch-clamp technique were carried out as described in detail previously (24). Perforated-patch studies included ∼225 μM amphotericin B in the pipette to permeabilize the patch membrane, and 50 mM NH₄⁺ to clamp pH_i to pH_o. For whole-cell studies, the main pipette solution (also used externally) contained (in mM) 130 TMACH₃SO₃, 2 MgCl₂, 2 EGTA, and 80 MES, titrated to pH 5.5 with ∼20 tetramethylammonium hydroxide (TMAOH). Bath solutions at pH 7.0 had (in mM) 90 TMACH₃SO₃, 3 CaCl₂, 1 EGTA, 100 N,N-Bis-(2-hydroxyethyl)-2-aminoethanesulfonic acid (BES), and 36–40 TMAOH. When pH was varied, the following buffers with pK_a near the desired pH were used: Homopipes for pH 4.6, MES for pH 5.5–6.0, BisTris for pH 6.5, Hepes for pH 7.5, and Tricine for pH 8.0. Experiments were done at 21 °C or at room temperature (20–25 °C). No leak correction has been applied to current records.

Patch-Clamp Studies of CLL and Healthy B Cells.

Cells were shipped frozen on dry ice. After thawing, cells were suspended in RPMI with serum, centrifuged at 1,100 × g for 5 min, and resuspended in a small volume (1–2 mL) of RPMI with serum.

Statistics.

Statistical analysis was carried out by Student’s unpaired t test. Further materials and methods are provided in SI Appendix.