Ablation of α2δ-1 inhibits cell-surface trafficking of endogenous N-type calcium channels in the pain pathway in vivo

Significance Neuronal N-type (CaV2.2) voltage-gated calcium channels are important at the first synapse in the pain pathway. In this study, we have characterized a knockin mouse containing CaV2.2 with an extracellular HA tag to determine the localization of CaV2.2 in primary afferent pain pathways. These endogenous channels have been visualized at the plasma membrane and rigorously quantified in vivo. We examined the effect of ablation of the calcium channel auxiliary subunit α2δ-1 (the target of gabapentinoids) on CaV2.2 distribution. We found preferential cell-surface localization of CaV2.2 in DRG nociceptor neuron cell bodies was lost, accompanied by a dramatic reduction at dorsal horn terminals, but no effect on distribution of other spinal cord synaptic markers.

The auxiliary α 2 δ calcium channel subunits play key roles in voltage-gated calcium channel function. Independent of this, α 2 δ-1 has also been suggested to be important for synaptogenesis. Using an epitope-tagged knockin mouse strategy, we examined the effect of α 2 δ-1 on Ca V 2.2 localization in the pain pathway in vivo, where Ca V 2.2 is important for nociceptive transmission and α 2 δ-1 plays a critical role in neuropathic pain. We find Ca V 2.2 is preferentially expressed on the plasma membrane of calcitonin gene-related peptide-positive small nociceptors. This is paralleled by strong presynaptic expression of Ca V 2.2 in the superficial spinal cord dorsal horn. EM-immunogold localization shows Ca V 2.2 predominantly in active zones of glomerular primary afferent terminals. Genetic ablation of α 2 δ-1 abolishes Ca V 2.2 cell-surface expression in dorsal root ganglion neurons and dramatically reduces dorsal horn expression. There was no effect of α 2 δ-1 knockout on other dorsal horn pre-and postsynaptic markers, indicating the primary afferent pathways are not otherwise affected by α 2 δ-1 ablation. calcium channel | primary afferent | auxiliary subunit | N-type | trafficking T he neuronal N-type voltage-gated calcium channel was first identified in primary afferent dorsal root ganglion (DRG) neurons (1,2). Toxins from the Conus marine snails, ω-conotoxin GVIA and ω-conotoxin MVIIC, are highly selective blockers of N-type channels (3,4) and have been instrumental in dissecting their function (5,6). A key role for N-type calcium channels was identified in primary afferent neurotransmission in the dorsal horn of the spinal cord, and these toxins were therefore pursued as a therapeutic target in the alleviation of chronic pain (7,8). Indeed, the peptide ziconotide (synthetic ω-conotoxin MVIIA) is licensed for intrathecal use in intractable pain conditions (9,10).
Despite the functional importance of N-type channels in the pain pathway, a major hindrance to the study of their distribution and trafficking, in this system and elsewhere, has been the paucity of antibodies recognizing this channel. Although previous studies have used anti-peptide antibodies to intracellular Ca V 2.2 epitopes (for example, refs. 11 and 12), these have not shown plasma membrane localization of the endogenous channel in neurons and have not been rigorously examined against knockout tissue. For this reason, we developed a Ca V 2.2 construct with an exofacial epitope tag to detect its cell-surface expression and trafficking (13). This channel is observed on the plasma membrane, when expressed in DRGs and other neurons (13)(14)(15). We took advantage of our finding that the presence of the epitope tag did not affect function (13) to generate a knockin (KI) mouse line containing the hemagglutinin (HA) tag in the same position in the Cacna1b gene. This has allowed us to examine the distribution of native Ca V 2.2 protein in the intact nervous system. N-type calcium channels are made up of the Ca V 2.2 poreforming α1-subunit, which associates with auxiliary α 2 δand β-subunits (16). Many studies have indicated that α 2 δ-subunits are important for the correct trafficking and physiological function of the channels (for a review, see ref. 17). A significant role for α 2 δ-1 in chronic neuropathic pain, which results from damage to peripheral sensory nerves, was identified as a result of two advances. First, it was shown that α 2 δ-1 mRNA and protein are strongly up-regulated in somatosensory neurons following nerve damage (18)(19)(20). Second, α 2 δ-1 was identified as the therapeutic target for the drugs gabapentin and pregabalin, which are used in neuropathic pain such as postherpetic neuralgia (21,22). Furthermore, α 2 δ-1 overexpression in mice resulted in a chronic painlike phenotype (23), whereas knockout of α 2 δ-1 caused a marked delay in the development of neuropathic mechanical hypersensitivity (24). However, it has not yet been possible to examine the effect of α 2 δ-1 on the trafficking of the relevant N-type channels in vivo.
Here we elucidate the cellular and subcellular localization of native Ca V 2.2 in neurons of the peripheral somatosensory nervous system. We reveal a dramatic effect of α 2 δ-1 ablation on Ca V 2.2 distribution, particularly in a key subset of nociceptive sensory neurons. In contrast to an early study of the subunit composition of N-type channels (16), which showed an ∼1:1 stoichiometry with α 2 δ-1, a more recent study suggested that α 2 δsubunits were only associated with less than 10% of digitoninsolubilized Ca V 2 channels (25), although it cannot be ruled out that they became dissociated during solubilization. However, the present study reinforces the essential nature of the auxiliary Significance Neuronal N-type (Ca V 2.2) voltage-gated calcium channels are important at the first synapse in the pain pathway. In this study, we have characterized a knockin mouse containing Ca V 2.2 with an extracellular HA tag to determine the localization of Ca V 2.2 in primary afferent pain pathways. These endogenous channels have been visualized at the plasma membrane and rigorously quantified in vivo. We examined the effect of ablation of the calcium channel auxiliary subunit α 2 δ-1 (the target of gabapentinoids) on Ca V 2.2 distribution. We found preferential cellsurface localization of Ca V 2.2 in DRG nociceptor neuron cell bodies was lost, accompanied by a dramatic reduction at dorsal horn terminals, but no effect on distribution of other spinal cord synaptic markers. α 2 δ-1 protein for cell-surface expression of endogenous Ca V 2.2, both in DRG neuronal cell bodies and in their presynaptic terminals. No effect of α 2 δ-1 loss was observed on other pre-and postsynaptic markers in the dorsal horn, despite a previous study implicating postsynaptic α 2 δ-1 in thrombospondin-mediated synaptogenesis (26). Our results therefore show that loss of synaptic Ca V 2.2 as a result of α 2 δ-1 ablation is due to a reduction of Ca V 2.2 trafficking to synapses, rather than synapse loss.

Results
Generation of Ca V 2.2_HA Knockin Mice. Mice containing a double-HA tag in constitutive exon 13 of the Cacna1b gene were generated in a C57BL/6 background, as described in Methods, such that every endogenous Ca V 2.2 contained the double-HA tag in the position previously ascertained not to affect channel function (13) (Fig. 1A). The presence of the HA tag was confirmed by PCR (Fig. 1B). We confirmed that the HA-tagged Ca V 2.2 protein is expressed in synaptosomes, since a 261-kDa band (the expected molecular mass of Ca V 2.2_HA) is recognized by anti-HA antibodies in Western blots of spinal cord tissue from Ca V 2.2_HA KI/KI , but not Ca V 2.2 WT/WT , mice (Fig. 1C).    (Fig. 1D). The properties of calcium channel currents in cultured DRG neurons from 10-to 12-wk-old Ca V 2.2_HA KI/KI mice were not altered compared with those from Ca V 2.2 WT/WT mice, both in terms of current density and voltage-dependent properties ( Fig.   1 E and F). We then examined whether Ca V 2.2_HA was detectable on the cell surface of cultured DRG neurons from Ca V 2.2_HA KI/KI mice (Fig. 1G). We found Ca V 2.2_HA to be present on the cell surface particularly of calcitonin gene-related peptide (CGRP)-positive peptidergic nociceptors, to a much greater extent than on isolectin-B4 (IB4)-positive nonpeptidergic nociceptors (56.8%, compared with 11.3%; Fig. 1 G and H). Furthermore, Ca V 2.2_HA was expressed on only a small  , Table S2.
Cell-Surface Expression of Ca V 2.2_HA in DRG Neurons in Vivo. In agreement with the results from cultured DRG neurons, we found that Ca V 2.2_HA was clearly present on the cell surface of DRG neuronal somata in sections of ganglia from 10-to 12-wk-old Ca V 2.2_HA KI/KI mice (Fig. 2 A, i-iv), and absent from Ca V 2.2 WT/WT mice (Fig. 2 A, v). We costained with markers of DRG neuronal subtypes, including CGRP (Fig. 2 A, i, ii, and v) and NF200 (Fig. 2 A, iii and iv). Analysis of the ratio of Ca V 2.2_HA at the cell perimeter, relative to its cytoplasmic staining, shows that plasma membrane Ca V 2.2_HA density is highest on the cell surface of small CGRP-positive DRG neurons (Fig. 2B). The small cell-surface Ca V 2.2_HA-positive DRG neurons were mainly NF200-negative (Fig. 2C). The absolute level of cytoplasmic staining of Ca V 2.2_HA was also negatively correlated with the size of DRG neurons (Fig. 2D), being higher in small-diameter neurons and in those which are CGRP-positive (Fig. 2D) and NF200-negative (Fig. 2E).  Fig. S2A). We found the level of α 2 δ-1 to be highest in CGRP-positive small DRG neurons (Fig. 3 A-C and SI Appendix, Fig. S2A). As expected, Ca V 2.2_HA KI/KI x α 2 δ-1 KO/KO DRG neurons show no staining for α 2 δ-1 above background (Fig. 3 A-C).
The effect of genetic ablation of α 2 δ-1 on Ca V 2.2_HA cell-surface expression was in general very marked (Fig. 3 D-F). We found that Ca V 2.2_HA was not concentrated on the cell surface in α 2 δ-1 KO/KO DRG neurons (Fig. 3D), and this was true across all subtypes of DRG neuron examined (Fig. 3 E and F). Furthermore, there was an increase in mean intracellular Ca V 2.2_HA intensity in DRG neurons from α 2 δ-1 KO/KO compared with α 2 δ-1 WT/WT mice, which was found in CGRP-positive DRG neurons (6.9% increase; Fig. 3G), and in both NF200-negative and NF200-positive DRG neurons (15.3 and 24.6% increase, respectively; Fig. 3H). The elevated intracellular Ca V 2.2_HA intensity in α 2 δ-1 KO/KO DRG neurons was also inversely correlated with cell size (SI Appendix, Fig. S2B).
Ca V 2.2_HA Is Localized in the Dorsal Horn of the Spinal Cord. Next, we examined the distribution of Ca V 2.2_HA in the spinal cord, and found strong immunoreactivity for the channel subunit in the dorsal horn (Fig. 4A). There was very little Ca V 2.2_HA in the ventral horn (Fig. 4A), and no specific staining in Ca V 2.2 WT/WT spinal cord (Fig. 4 A, i). Taking regions of interest (ROIs) perpendicular to the pial layer ( Fig. 4 A, ii), we found that within the dorsal horn, Ca V 2.2_HA was most abundant in superficial laminae I and II (Fig. 4B). Here Ca V 2.2_HA shares topographic distribution with both the presynaptic markers CGRP, which is present in peptidergic primary afferent terminals in laminae I and II-outer ( Fig. 4C and SI Appendix, Fig. S3A), and with IB4, which is present in nonpeptidergic terminals, mainly in lamina IIinner ( Fig. 4D and SI Appendix, Fig. S3B). Ca V 2.2_HA was also associated with a postsynaptic marker of excitatory synapses, Homer (Fig. 4E).
Ablation of α 2 δ-1 Reduces Ca V 2.2_HA in the Dorsal Horn Without Effect on Other Synaptic Markers. The distribution of Ca V 2.2_HA in the dorsal horn was markedly reduced in α 2 δ-1 KO/KO mice (Fig. 4 F-H), particularly in the superficial layers (Fig. 4I). Fol-lowing subtraction of nonspecific signal found in wild-type Ca V 2.2 sections (Fig. 4B), the reduction in Ca V 2.2_HA was 72.7, 65.9, 64.6, and 44.7% in layers I, II-outer, II-inner, and III, respectively (Fig. 4I). This decrease provides clear evidence for the essential role of α 2 δ-1 for Ca V 2.2 trafficking to the primary afferent presynaptic terminals. In contrast, in the deeper layers of the dorsal horn (laminae IV and V), there was no effect of the ablation of α 2 δ-1 on the low level of Ca V 2.2_HA present (Fig. 4I).
Next, we investigated whether the α 2 δ-1-mediated loss of Ca V 2.2_HA in the dorsal horn was concomitant with a reduction in density or distribution of synaptic markers, since α 2 δ-1 has also been implicated in synaptogenesis (26). In contrast to the marked reduction in Ca V 2.2_HA in the absence of α 2 δ-1 (Fig.  4I), there was no effect of α 2 δ-1 ablation on the overall immunostaining intensity or distribution in the dorsal horn of three primary afferent presynaptic markers, CGRP (Fig. 4J), IB4 (Fig.  4K), and vesicular glutamate transporter-2 (vGlut2) (Fig. 4L), and no effect on postsynaptic Homer immunostaining (Fig. 4M).

Dorsal Rhizotomy Reduces Ca V 2.2_HA in the Dorsal Horn of the Spinal
Cord. In light of the marked reduction in Ca V 2.2_HA, without loss of synaptic markers, in the dorsal horn of α 2 δ-1 KO/KO mice (Fig. 4I), we wished to examine further the extent of its origin in presynaptic primary afferent terminals. To investigate this, we performed unilateral dorsal rhizotomy (Fig. 5A). This resulted in a significant reduction of Ca V 2.2_HA in the ipsilateral dorsal horn (Fig. 5 B-D). In the central ROI, the reduction was 52.7% in the superficial layers I and II, and there was also a substantial depletion (by 44.7%) in layers III to V (Fig. 5D). Rhizotomy is generally found to be incomplete, as longitudinal fibers remain intact (27). To determine the extent of the rhizotomy, we also examined the level of CGRP, as a marker of loss of presynaptic peptidergic afferents (27). A very similar extensive reduction of CGRP was observed, by 53.1% in layers I and II and 58.6% in layers III to V (Fig. 5 E and F). The correspondence between the reduction of Ca V 2.2_HA and that of CGRP, whose origin is entirely presynaptic in the dorsal horn, confirms the mainly presynaptic localization of the Ca V 2.2_HA signal in this region. Following dorsal rhizotomy, there was also a 20.7% decrease of α 2 δ-1 in central laminae I and II (SI Appendix, Fig. S4), which is expressed both in primary afferents and in intrinsic neurons (20). In contrast, there is no reduction in the NPY signal in the same region (SI Appendix, Fig. S4), this peptide being expressed mainly by dorsal horn interneurons (for a review, see ref. 28).
Ca V 2.2_HA Subcellular Localization in the Spinal Cord: Effect of α 2 δ-1 Ablation. At higher resolution, we observed that Ca V 2.2_HA, present in the superficial dorsal horn laminae, was distributed in rosette structures consisting of Ca V 2.2_HA puncta surrounding a central core containing vGlut2 and often (but not always) associated with either CGRP (SI Appendix, Fig. S5 A and B) or IB4 (SI Appendix, Fig. S5 C and D), resembling glomerular synapses (29).
To improve resolution of these structures, we then obtained superresolution Airyscan images of Ca V 2.2_HA together with vGlut2 and Homer in regions of the dorsal horn in both α 2 δ-1 WT/WT (Fig. 6A) and α 2 δ-1 KO/KO mice (Fig. 6B). The rosette-shaped clusters of Ca V 2.2_HA consisted of groups of four or five puncta (Fig. 6C). These puncta may each correspond to individual active zones of primary afferent terminal glomerular synapses, because they are usually organized around a central core containing vGlut2, and also frequently apposed to the postsynaptic marker Homer (Fig. 6C).
We found the density of Ca V 2.2_HA was markedly reduced in α 2 δ-1 KO/KO dorsal horn (Fig. 6 B and C), and we quantified the effect on several parameters associated with Ca V 2.2_HA puncta (for a method, see SI Appendix, Fig. S6). The density of Ca V 2.2_HA was reduced in individual clusters of puncta in α 2 δ-1 KO/KO dorsal horn, by 47.7% (Fig. 6D), but the cluster areas were not significantly affected (Fig. 6E). In contrast, neither the area nor the intensity of vGlut2 or Homer clusters was affected by loss of α 2 δ-1 (Fig. 6 D and E). In estimating the pairwise association between Ca V 2.2_HA and Homer (Fig. 6F), or Ca V 2.2_HA and vGlut2 (Fig. 6G), we found that the intensity of vGlut2 and Homer in these associated clusters was not affected in α 2 δ-1 KO/KO dorsal horn (Fig. 6  F and G). However, as expected, the intensity of Ca V 2.2_HA in the associated clusters was reduced by 50.0% for Ca V 2.2_HA puncta overlapping with Homer (Fig. 6F), and by 50.7% for those overlapping with vGlut2 (Fig. 6G). Subcellular Localization of Ca V 2.2_HA. To determine the subcellular localization of the Ca V 2.2_HA channels, we used preembedding immunogold labeling. For electron microscopic investigation, tissue blocks were taken from the dorsal horn of the spinal cord. Immunoreactivity for Ca V 2.2_HA was predominantly found in presynaptic elements, namely on axon terminals of presumed primary afferents (Fig. 7 A-C). Single or small clusters of immunogold particles were mainly localized to the active zone of boutons, including multiple active zones on individual glomerular boutons (Fig. 7 B and C), and also appeared at the edge of presynaptic membrane specializations (Fig. 7 A-C) and along the extrasynaptic plasma membrane (Fig. 7 A-C) of axon terminals making asymmetrical putative glutamatergic synapses with dendritic shafts and spines of postsynaptic neurons. The specificity of the immunolabeling was confirmed by the absence of immunoreactivity for Ca V 2.2_HA in tissues obtained from control animals (Fig. 7D).  Scatter plots of Ca V 2.2_HA intensity (with blue mean ± SEM) for data from C, in superficial laminae I and II and in laminae III to V, contralateral (black circles) and ipsilateral (red circles) to rhizotomy. ****P < 0.0001, *P = 0.014 (paired t test). (E) Plot profile of CGRP intensity (mean ± SEM of 15 sections, normalized to the average contralateral intensity between 4 and 24 μm) in dorsal horn ROIs, contralateral (black line) and ipsilateral (red line) to rhizotomy. (F) Scatter plots of CGRP intensity (with blue mean ± SEM) for data from E, in superficial laminae I and II and in laminae III to V, contralateral (black circles) and ipsilateral (red circles) to rhizotomy. ****P < 0.0001 (paired t test).

Discussion
In this study, we have been able to visualize native N-type Ca V 2.2 channels on the cell surface of neurons in vivo. We have concentrated here on the primary afferent neuronal pathway, because of the importance of Ca V 2.2 in synaptic transmission in this system and its therapeutic importance as a drug target (7,30). We show that Ca V 2.2_HA is very strongly expressed on the cell surface, particularly of CGRP-positive small DRG neurons, and this is recapitulated in DRG neurons in culture. In contrast, transcriptional profiling found Cacna1b mRNA to be present in similar amounts in IB4-positive and IB4-negative nociceptors, the latter group including CGRP-positive DRG neurons (31). This would agree with the high intracellular Ca V 2.2_HA we found in both CGRP-positive and CGRP-negative small DRG neurons. The localization of Ca V 2.2_HA in DRG neurons is paralleled by striking expression of Ca V 2.2_HA in the dorsal horn of the spinal cord, predominantly in laminae I and II. Here the presynaptic Ca V 2.2_HA puncta are associated with the primary afferent markers CGRP, vGlut2, and IB4, present in glomerular primary afferent presynaptic terminals as described previously (29). The Ca V 2.2_HA puncta are also adjacent to puncta containing the postsynaptic density protein Homer. The presynaptic localization of Ca V 2.2_HA in primary afferents is confirmed through their ablation by dorsal rhizotomy. Furthermore, from the high-resolution immunoelectron-microscopic localization of Ca V 2.2_HA, we confirm that these rosette structures formed by the Ca V 2.2_HA puncta are likely to represent Ca V 2.2_HA in active zones of individual glomerular terminals. The α 2 δ-1 auxiliary subunit has been shown to be important for calcium channel trafficking in expression systems (13). It plays a major role in pain pathways and is up-regulated following neuropathic injury (17)(18)(19)(20)23). Furthermore, knockout of α 2 δ-1 caused a marked delay in the development of neuropathic mechanical hypersensitivity (24), and overexpression of α 2 δ-1 mimics features of neuropathic injury (23). In rats, α 2 δ-1 is expressed in all DRG neurons with highest expression in small neurons (20), and this distribution is confirmed here, in mice. However, until now it has not been possible to examine the effect of α 2 δ-1 on the trafficking of the relevant endogenous N-type channels in vivo.
Our results using Ca V 2.2_HA KI/KI mice crossed with α 2 δ-1 KO/KO mice, in which α 2 δ-1 is globally ablated, highlight the essential role of α 2 δ-1 in directing Ca V 2.2_HA to the cell surface in DRG neurons and in targeting Ca V 2.2_HA to presynaptic terminals in the dorsal horn. Accompanying the complete loss of DRG neuronal cell-surface Ca V 2.2_HA, there was also a significant increase in cytoplasmic Ca V 2.2_HA in CGRP-positive α 2 δ-1 KO/KO DRG neurons, indicating a defect in cell-surface trafficking.
The calcium currents in DRG neuronal somata in culture are found to be composed of between 20 and 50% N-type current, depending on the species, developmental stage, culture conditions, and subtype of DRG neuron examined (24,(32)(33)(34)(35). One comprehensive study showed the proportion of N-type current was about 40% in cultured mouse DRG neurons with a diameter of less than 30 μm, and 20% in those larger than 30 μm (35), which is in agreement with the differential distribution of Ca V 2.2_HA found here in small DRG neurons. We found previously that in cultured DRG neurons from α 2 δ-1 knockout mice the calcium channel current was only reduced by about 30% compared with wild-type DRG neurons, and the N-type current was reduced proportionately (24), which is in contrast to the marked effects of α 2 δ-1 knockout on Ca V 2.2_HA localization described here. It is highly likely that even short-term cultured DRG neurons do not fully represent the in vivo situation, and that rapid changes occur in cell-surface expression of receptors and channels when cells are enzymatically dissociated and maintained in culture, allowing neurite outgrowth (36). Since evoked synaptic currents in laminae I and II are 74% N-type (37), there is likely to be a differential synaptic localization of these channels in vivo.
It has been found that there are other synaptic roles for α 2 δsubunits unrelated to calcium channel function; for example, an association of the extreme C terminus of α 2 δ-1 with NMDA receptors has been identified (38). Furthermore, postsynaptic α 2 δ-1 has been implicated in central neurons as the binding partner of thrombospondins to promote synaptogenesis induced by this secreted protein family, independent of its role as a calcium channel subunit (26,39). Thrombospondins alone promote the formation of silent synapses, lacking postsynaptic elements (40). However, we did not detect robust binding of thrombospondin-4 to α 2 δ-1 (41). By contrast, in cultured hippocampal neurons, neuroligin was also identified as a binding partner of thrombospondins mediating an increase in the rate of synaptogenesis (42).
Both presynaptic α 2 δ-3 (43) and α 2 δ-4 (44) have also been implicated in determining synaptic morphology in the auditory system and retina, respectively, although in these cases the synaptic abnormalities resulting from knockout of the respective α 2 δ-subunits are likely related to calcium channel dysfunction. In the present study, despite the effect of global ablation of α 2 δ-1, which strongly disrupted Ca V 2.2_HA cell-surface localization, particularly of CGRP-positive small DRG neurons, and markedly reduced presynaptic terminal localization of Ca V 2.2_HA in the dorsal horn of the spinal cord, we did not observe any reduction in other presynaptic markers of these primary afferents, CGRP, vGlut2, and IB4, or the postsynaptic marker, Homer. At the level of individual synapses, we did not find a reduction in area of Ca V 2.2_HA-positive puncta clusters, but there was a very clear reduction in intensity of Ca V 2.2 in each cluster, in the absence of α 2 δ-1. This result suggests that, if these puncta represent presynaptic active zones in primary afferent glomerular synapses, α 2 δ-1 has not affected the density of synapses in the dorsal horn, despite a large reduction in presynaptic Ca V 2.2_HA intensity. However, whether there are changes in synaptic morphology will require more detailed examination at the EM level in the future.

Methods
Generation of Ca V 2.2_HA Epitope-Tagged Knockin Mice. The Ca V 2.2_HA mouse line was generated by Taconic Artemis in the C57BL/6 background by homologous recombination with the targeting vector, which included the genomic region around exon 13 of the Cacna1b gene from clones of a C57BL/6J RPCIB-731 BAC library into which the sequence coding for the 2× HA tag was cloned. The targeting vector also carried the puromycin resistance gene (PuroR) as a positive-selection marker in intron 13 between two Flipper recombination sites and the negative-selection marker thymidine kinase outside the homologous regions. The targeting vector was linearized and transfected into embryonic stem cells. The homologous recombinant clones were isolated by positive and negative selection and injected into blastocysts from BALB/c. Highly chimeric mice were crossed with C57BL/6, and transmission to the germ line was confirmed by black offspring. The positive selective marker was removed by Flipper recombinase after crossing the first generation of knockin mice with Flp deleter transgenic mice. Subsequent backcrossing with wild-type C57BL/6 mice allowed us to select mice without the Flipper transgene and only the 2× HA tag insertion in exon 13. Genotyping PCR was performed with the primers forward, 5′-CACACCAGCATACATGCTCG-3′ and reverse, 5′-TCCAGCCTCACATGCTGC-3′, that bind to the intronic sequences just before and after exon 13 to generate