Absence of venous valves in mice lacking Connexin37

Mice

Cx37−/− (Gja4−/−) (Simon et al., 1997) mice were maintained on a C57BL/6 background and genotyped by PCR using previously published methods (Kanady et al., 2011). Animal protocols were approved by the IACUC Committee at the University of Arizona (Tucson, AZ).

Antibodies

Primary antibodies used for immunostaining were as follows: rabbit antibodies to Cx32 (71-0600, Invitrogen), Cx37 (Simon et al., 2006), Cx40 (Gabriels and Paul, 1998), Cx43 (C6219, Sigma), Cx43 (71-0700, Invitrogen), Cx47 (364700, Invitrogen), Prox1 (11-002, AngioBio), Prox1 (ab11941, Abcam), von Willebrand Factor, vWF (A0082, Dako); mouse antibodies to Cx43 (Cx43 IF1, a gift from Paul Lampe); rat antibodies to CD31 (550274, BD Biosciences), CD144 (555289, BD Biosciences); goat antibodies to EphB4 (AF446, R&D Systems), Foxc2 (ab5060, Abcam), VEGFR3 (AF743, R&D Systems). AffiniPure minimal cross reactivity secondary antibodies (conjugated to Alexa 488, Alex 555, Alexa 647, Cy3, Cy5, or Dylight 649) and unlabeled Fab fragments were from Jackson Immunoresearch. For some experiments, Cx37 antibody was directly conjugated to Alexa 555 using the Alexa Fluor 555 Protein Labeling Kit (A20174, Invitrogen) and Cx43 antibody (C6219, Sigma) was directly conjugated to Alexa 488 using the APEX Alexa Fluor 488 Antibody Labeling Kit (A10468, Invitrogen).

Immunostaining

Tissues were frozen unfixed in Tissue-tek O.C.T. and sectioned at 10 µm. Sections were fixed in acetone at −20°C for 10 minutes, blocked in a solution containing PBS, 4% fish skin gelatin, 1% donkey serum, 0.25% Triton X-100, and incubated with primary antibodies for 1.5 to 3 hours at room temperature or overnight at 4°C. Sections were washed with PBS containing 0.25% Triton X-100 and then incubated with secondary antibodies for 30–45 minutes. For experiments in which Alexa 555-Cx37 antibody or Alexa 488-Cx43 antibody (directly conjugated rabbit antibodies) were used for double labeling along with another unlabeled rabbit primary antibody, the unlabeled rabbit antibody was incubated on sections first, followed by the secondary antibody and a blocking step with 5% normal rabbit serum, before the directly conjugated antibody was applied. After washing, sections were mounted in Mowiol 40–88 (Aldrich) containing 1,4-diazobicyclo-(2,2,2)-octane and viewed with an Olympus BX51 fluorescence microscope fitted with a Photometrics CoolSnap ES2 camera or viewed with a Zeiss LSM 510 confocal microscope.

Saphenous artery/vein pair ligation

Mice were anesthetized with 1.5% isoflurane in O₂ during surgery and injected with buprenorphine (0.1 mg/kg) for analgesia prior to the operation. An incision was made in the left hindlimb skin and the saphenous artery-vein pair was exposed and then ligated at the knee, just above the sapheno-femoral junction, and again near the ankle, using 6-0 silk suture. The intervening saphenous artery-vein pair segment, with disrupted blood flow, was left in place and the incision was closed. For sham surgery controls, the procedure was the same except that the ligation was omitted or, in some experiments, only the surgical incision was done (for the same amount of time), without exposure of the blood vessels. 8, 24, or 72 hrs after surgery, the mice were sacrificed and the saphenous artery-vein pair was collected and frozen unfixed in Tissue-tek O.C.T. and sectioned longitudinally at 10 µm section thickness for immunostaining. In some experiments, the femoral and iliac artery-vein pairs were also also collected.

Blood pressure measurements

Blood pressures were measured with a Hatteras Instruments multi channel blood pressure tail cuff analysis system (MC4000). Measurements were recorded on four days for each mouse with 15 measurements recorded in each individual run. 9 wild-type and 9 Cx37−/− mice were tested.

In valved veins, Cx37, Cx43, and Cx47 are found exclusively at valves

Cryosections containing venous valves from mouse saphenous, brachial, iliac, or azygos veins were immunostained with antibodies against Cx37, Cx43, and Cx47, Cx family members previously found to be highly enriched at lymphatic valves (Kanady et al., 2011). Because Foxc2, which may be a direct regulator of Cx37 gene (Gja4) expression, is enriched at venous valves as well as lymphatic valves, we hypothesized that Cx37 would be prevalent at venous valves. Fig. 1 shows that Cx37 was indeed highly expressed at venous valves but was undetectable in the non-valve venous endothelium (see also Figs. 2–4 and Fig. S1 in the supplemental material). Control experiments (CD31, CD144, EphB4, vWF immunostaining) confirmed that the vascular endothelium was intact throughout the vein following the collection and processing of the vein samples (Fig. 1F). Cx43 and Cx47 were also found exclusively at the valves in these veins (Figs. 2–4 and Fig. S1 in the supplemental material). Cx40, a Cx that is highly abundant in arterial endothelium (Bruzzone et al., 1993), was not detected either at venous valves or in non-valve regions of the vein, nor was Cx32, a Cx family member reportedly expressed in mouse vena cava (Okamoto et al., 2009) (see Fig. S2 in the supplemental material). Within the venous valves, Cx37, Cx43, Cx47 immunosignals were concentrated at cell-cell interfaces between the valve endothelial cells.

Cxs are differentially expressed on the upstream and downstream faces of venous valve leaflets

Double labeling showed that Cx37 and Cx43 were remarkably differentially localized within the valve leaflets, showing little or no colocalization (Fig. 2C,E,F,H,I; See also Fig. 3D and E; and Fig. S1 in the supplemental material). Venous valve leaflets are comprised of two endothelial layers which are separated by a layer of extracellular matrix. From the appearance of the Cx double labeling, we hypothesized that Cx37 and Cx43 were expressed on different sides of the valve leaflet. Consistent with this idea, confocal z-stack imaging of sections in which the valve leaflet was relatively flat (en face) and intact (Fig. 2E), showed a clear separation of Cx37 and Cx43 immunosignals in the z-axis. The signal separation in the x-z and y-z projections in Fig. 2E was not perfect, however this could be explained if the leaflet was not entirely flat. In addition, although the majority of the Cx staining in the samples originated from the valve endothelial cells, Cx43-expressing leukocytes occasionally adhere to valve leaflets and can contribute to the Cx43 signal in certain spots. To look at the distribution of Cx37 and Cx43 in a different way, we also immunostained several transversely sectioned valve leaflets in which the leaflet is viewed essentially end-on (Fig. 2F–I). These experiments showed again a distinct separation of Cx37 and Cx43 on the two sides of the valve leaflet. Because the orientation and topology of the en face as well as transversely sectioned venous valves was known, we were able to conclude that Cx43 is highly localized on the upstream side of the valve leaflet, whereas Cx37 is localized on the downstream side of the leaflet (see schematic illustration in Fig. 2J).

In contrast to Cx37 and Cx43, Cx47 was consistently found to be highly restricted to a small subset of cells in the venous valve (Figure 3H,I and J). Interestingly, there were often two types of Cx47 staining patterns within the same localized area of the valve: a weaker punctate signal that generally outlined a few endothelial cells and a second, much more intense, line-like signal that did not outline entire cells. The spatial relationship of Cx47 expression in the valve to that of Cx37 and Cx43 expression was complex (Fig. 4). Double labeling clearly showed that the weaker punctate Cx47 signal colocalized with Cx43 (Fig. 4K). However, the intense line-like Cx47 signal did not colocalize with Cx43 (Fig. 4N). Rather, this intense Cx47 signal sometimes overlapped with Cx37 (Fig. 4G and H), while in other cases the Cx47 and Cx37 signals were distinct (Fig. 4C and D). These results show that venous valves are highly enriched for at least three Cxs which are differentially expressed on the two sides of the valve leaflets and, in addition, differ in their breadth of expression, spatially, throughout the leaflet face.

In addition to valved veins, we also surveyed several veins (i.e. superior vena cava, inferior vena cava, mesenteric veins draining the small intestine, superior mesenteric vein leading into the portal vein), which typically do not contain valves (see Fig. S3 in the supplemental material). Cx37 and Cx43, but not Cx47 or Cx40, were detected in the thoracic portion of the inferior vena cava (Fig. S3 A–F) but were not detected in the abdominal portion of this vessel (not shown). Weak Cx37 expression, but not Cx43, Cx47 or Cx40 expression was detected in the superior vena cava (not shown). By comparison, in the rat, Inai and Shibata found heterogeneous expression of Cx37, Cx40, and Cx43 in vena cava with the relative levels of Cxs varying depending on the position (e.g. superior versus inferior vena cava) (Inai and Shibata, 2009). In contrast to the findings of Okamoto et al. (Okamoto et al., 2009), we did not detect Cx32 in any of the vena cava regions tested even though the Cx32 antibody successfully labeled positive control tissues (nerves and hepatocytes) (Fig. S2H and I). Although most sections of the mesenteric veins draining the small intestine showed no Cx expression, some portions of these vessels exhibited very low, patchy Cx37 staining. Interestingly, however, at points of convergence between adjacent mesenteric veins, strong Cx37 upregulation was consistently observed in the venous endothelium (Fig. S3 M–O). Finally, only Cx37 was detected in the much larger superior mesenteric vein leading to the portal vein (Fig. S3 G–L), whereas both Cx37 and Cx40 were previously reported in the endothelium of mouse portal vein (Shiojiri et al., 2006). Thus, while Cx expression in valved veins is tightly restricted to the valve region, these results show that in non-valved veins, Cx37 and Cx43 (and Cx40 in some studies) can be present in the general venous endothelium, depending on the particular vein. The expression of Cxs in valved and non-valved veins examined in this study is summarized in Table S1 (see supplemental material).

Cx37 is required for the presence of venous valves

Cx37−/− mice (Simon et al., 1997) were used to investigate the role of Cx37 in venous valve formation and/or maintenance. Mice lacking Cx37 are viable, but are deficient in lymphatic valves (Kanady et al., 2011). However, the status of their venous valves had not been determined. To address this issue, longitudinal serial sections of saphenous or brachial vein from adult wild-type or Cx37−/− mice were immunostained for Cx37 (for wild-type samples), Cx43, and Prox1 to aid in the identification of venous valves. In agreement with previous reports (Norrmén et al., 2011; Srinivasan and Oliver, 2011), we found that the transcription factors Prox1 and Foxc2 are highly enriched at venous valves (Fig. 1C, E, and I; Fig. 6H). To make certain that venous valves could be detected if present in Cx37−/− veins, we confirmed that wild-type venous valves were readily imaged with a combined Cx43 and Prox1 immunostaining protocol. Vein identification was checked on adjacent sections by EphB4 immunostaining (specific for veins) or by double labeling sections for vWF and VEGFR3. In the mouse, veins are positive for vWF and negative for VEGFR3, whereas lymphatics are negative for vWF and positive for VEGFR3. Using this approach, we detected valves in 18 out of 19 serially sectioned wild-type veins (Fig. 5A and H). Remarkably, however, we did not find a single valve in 9 serially sectioned Cx37−/− veins (Fig. 5B,C, and H). Moreover, direct in vivo examination of exposed, intact brachial and iliac veins with a stereomicroscope confirmed the presence of valves in wild-type veins and the absence of valves in Cx37−/− veins (3 wild-type mice, 2 Cx37−/− mice, 4 veins examined per mouse) (Fig. 5 D–H). These data, summarized in Fig. 5H, strongly show that Cx37 is required for the presence of venous valves in the mouse.

Despite the absence of venous valves (and a deficiency in lymphatic valves), Cx37−/− mice are overtly healthy, although they do exhibit lymph reflux (Kanady et al., 2011). Systolic blood pressures, measured by the tail-cuff method, were not significantly different between 9 wild-type and 9 Cx37−/− mice (WT: 103.3 +/− 3.8 versus Cx37−/−: 104.5 +/− 3.0 mmHg; P=0.81, t-test). Diastolic pressure, mean arterial pressure, and pulse rate were also not significantly altered in the Cx37−/− mice (not shown).

Cx37 in venous valves is maintained after a prolonged disruption of venous blood flow

Cx expression in blood vessels is known to be affected by conditions that produce turbulent or disturbed blood flow (Gabriels and Paul, 1998). In addition, oscillatory (versus laminar) fluid shear stress influences Cx37 and Cx43 expression in cultured lymphatic endothelial cells (Sabine et al., 2012). To determine if the highly localized expression of Cxs at venous valves is dependent upon normal blood flow through the vein, we unilaterally ligated the saphenous artery-vein pair at the knee and again near the ankle, leaving the intervening portion of the saphenous vein in place with severely disrupted blood flow. After 8, 24, or 72 hrs, the saphenous artery-vein pairs from the ligated limb and the unligated control (contralateral) side were collected, sectioned longitudinally, and immunostained for Cx37, Cx43, and Cx47. Cxs have a very short half-life (1–5 hrs), so preexisting Cxs should turnover during these intervals (Laird, 2006). However, Cx37 expression was consistently maintained at venous valves throughout these periods of disrupted flow, although the Cx37 pattern appeared somewhat disorganized compared to controls (Fig. 6A and B). Cx43 expression in the valves of ligated veins, was more variable but generally appeared decreased relative to control valves. At 8 hrs, there was still some Cx47 expression at valves, but by 24 hrs Cx47 was virtually absent from the ligated veins. In addition to Cx37, Prox1 and Foxc2 expression was maintained at venous valves in the ligated veins (see Fig. 6H for Foxc2). These experiments indicate that Cx37 expression at venous valves persists during a prolonged disruption of venous blood flow, whereas Cx43 and Cx47 expression may be more dependent on normal laminar fluid flow.

Surgical incision of the lower limb skin, without vessel ligation, dramatically induces Cx43 expression in the non-valve endothelium of the saphenous vein

The saphenous artery-vein ligation experiments described above also showed that Cx43 expression was strikingly induced in the non-valve endothelium of the saphenous vein from the ligated limb (Fig. 6D and E; Fig. 7A). Although Cx43 induction was already evident at 8 hrs post-ligation, Cx43 levels were further increased by 24 hrs post-ligation. In contrast, Cx47 and Cx40 were not induced in the non-valve endothelium, and only a very small upregulation of Cx37 was observed at higher magnification in some areas. Cx43 was not induced in the vein from the contralateral, control limb, thus ruling out a global effect. We tested the idea that Prox1 or Foxc2 expression might be induced everywhere in the vein following ligation and therefore be responsible for the induction of Cx43 expression, however both of these transcription factors remained localized to the valve (see Fig. 6H for Foxc2).

Surprisingly, sham surgery control experiments revealed that surgical incision of the lower limb skin alone, without subsequent blood vessel ligation, was sufficient to induce high Cx43 expression in the non-valve endothelium (Fig. 7C). In contrast, at the valve itself, Cx43 was not upregulated after skin incision, rather, Cx43 expression tended to decrease at the valve. To get a better sense of how localized was the effect of surgical incision on Cx43 expression, we also looked at the femoral and iliac veins in the same incised limb. There was some induction of Cx43 expression in the femoral vein, especially in the part of the femoral vein that was closest to the saphenous vein, but the Cx43 expression was not as high as in the saphenous vein. In the iliac vein (more distant from the incision), Cx43 upregulation was not detected. Thus, the effect of a skin incision on Cx43 expression was quite localized, affecting only veins that were closest to the incision. This surgical incision (wounding) effect was highly specific for Cx43 (Fig. 7C and D). As with the ligations, only a very small induction of Cx37 expression was observed in some areas of the vein when viewed at higher magnification. The effect of skin wounding on Cx43 expression was transient, as 72 hrs after a surgical incision of the lower limb skin, Cx43 expression in the non-valve endothelium of the saphenous vein was largely gone (Fig. 7G). These results indicate that although the non-valve endothelium of superficial veins are seemingly devoid of Cxs normally, wounding of the overlying skin can fundamentally alter this state, at least transiently. Moreover, injury to the skin may lead to the local release of signals that specifically induce Cx43 in the nearby non-valve venous endothelium.