Background

Several aquaporin- (AQP) type water channels are expressed in kidney tubules and microvessels, including AQP1 in proximal tubule, thin descending limb of Henle and vasa recta, AQP2 in collecting duct apical membrane, and AQP3 and AQP4 in collecting duct basolateral membrane. Mice deficient in these aquaporins have distinct phenotypic abnormalities. AQP1 null mice are polyuria and unable to generate a concentrated urine after water deprivation. AQP2-T126M mutant mice and AQP3 null mice manifest nephrogenic diabetes insipidus (NDI) with severe polyuria, whereas AQP4 null mice have only a mild defect in maximal urinary concentrating ability. We reasoned that these mice could serve as useful models for gene replacement because of their predictable and unambiguous phenotypes.

Methods

In an initial feasibility study, an adenovirus directing the expression of AQP1 was introduced into AQP1 null mice by intravenous infusion.

Results

At 1 week after adenovirus infusion, AQP1 was seen in many proximal tubules and microvessels. Compared with untreated null mice, the treated mice were able to partially concentrate their urine and lost less weight after water deprivation. However, AQP1 transgene expression and functional correction were lost over 3–5 weeks.

Conclusion

Although there remain many technical problems to overcome, aquaporin gene replacement has potential applications in hereditary and acquired NDI, and in the transient modulation of renal fluid conservation.

INTRODUCTION

Gene delivery has been proposed to have potential applications in the therapy of hereditary and acquired diseases of the kidney including polycystic disease, nephrogenic diabetes insipidus, and tumors^1,2,3,4,5,6. Reporter gene expression has been studied following viral and non-viral gene delivery by renal artery, ureteral, and direct intrarenal injection^7,8,9,10. A single renal artery injection of adenovirus produced transient expression of the -galactosidase reporter in proximal tubule epithelial cells⁷. A similar study using adenovirus and renal artery delivery gave transient -galactosidase gene expression mainly in endothelial cells⁶. Retrograde ureteral injection of adenovirus gave transient reporter gene expression in papillary and medullary tubular cells⁷.

In a more recent study, delivery of the carbonic anhydrase II (CAII) gene in CAII-deficient mice by retrograde injection of a cationic liposome-CAII gene complex gave expression in tubular cells of the outer medulla and corticomedullary junction¹¹. Gene and transcript expression were highest at day 3 after treatment and remained detectable at 1 month. The CAII-deficient mice showed improved urinary acidification after NH₄Cl administration at 3 weeks but not at 6 weeks after gene delivery. To explore cardiac and renal protective effects of kallikrein gene delivery in chronic renal failure, Wolf et al.¹² delivered adenovirus carrying the human tissue kallikrein cDNA into rats after 5/6 reduction of renal mass. Kallikrein gene delivery significantly decreased total urinary protein and albumin excretion and increased levels of urinary kinin, nitrite/nitrate, and cGMP. Kallikrein gene transfer reduced glomerular sclerotic lesions, tubular damage, lumenal protein cast accumulation, and interstitial inflammation in the kidney, as well as myocardial hypertrophy.

Our laboratory recently generated a series of transgenic mice deficient in the major renal water channels, AQP1 through AQP4, as well as extrarenal water channels. Based on the predictable and unambiguous renal abnormalities in these mice, we have begun to use these mice as hosts to explore the feasibility of functional aquaporin gene delivery.

PHENOTYPE OF TRANSGENIC AQUAPORIN KNOCKOUT MICE

Distinct phenotype abnormalities in kidney and extrarenal organs have been identified in aquaporin null mice^13,14,15. Deletion of AQP1 produced a severe defect in urinary concentrating ability resulting in profound fluid loss when mice were deprived of water¹⁶. Urine osmolality remained in the range 600–700 mOsm even after a 36-hour water deprivation and administration of the vasopressin agonist DDAVP. Mechanistic analysis revealed that the urinary concentrating defect results from a combination of defective proximal tubule fluid absorption¹⁷, and defective countercurrent exchange because of low water permeability in thin descending limb of Henle¹⁸ and outer medullary descending vasa recta¹⁹. The decreased active fluid absorption in proximal tubule after AQP1 deletion supports the general paradigm that high epithelial cell water permeability is required for active, near-isosmolar fluid transport. Mice lacking AQP3, which is expressed mainly at the basolateral membrane of cortical collecting duct principal cells, manifest nephrogenic diabetes insipidus with polyuria, polydipsia, and urinary hypoosmolality²⁰. The decreased basolateral membrane water permeability impairs the osmotic extraction of water from the collecting duct lumen. Mice lacking AQP4, which is expressed mainly in inner medullary collecting duct, have a mildly impaired maximal urinary concentrating ability²¹ despite 4-fold decreased transepithelial water permeability²². In addition, double knockout mice lacking pairs of aquaporins (AQP1/AQP3, AQP1/AQP4, and AQP3/AQP4) have been generated and characterized. Recently, a mouse knock-in model of human NDI was created by targeted gene replacement in which the human disease-causing mutation T126M was introduced²³. The homozygous mutant mice express the AQP2-T126M protein but manifest severe polyuria and dehydration resulting in early neonatal death.

Several interesting extrarenal phenotype abnormalities have been documented in aquaporin null mice. Like the proximal tubule, the salivary gland requires active transepithelial fluid secretion to generate a near-isomolar primary saliva. Deletion of AQP5, the apical membrane water channel in salivary gland acinar epithelia, results in defective saliva production²⁴. Whereas wild-type mice produce large amounts of clear, non-viscous saliva following pilocarpine injection, AQP5 knockout mice produce relatively little viscous fluid. The saliva was hyperosmolar and hypernatremic, suggesting that AQP5 null mice are able to pump salt actively into the acinar lumen, but that water permeability is too low to permit osmotic equilibration. In brain, the most strongly expressed aquaporin is AQP4, which is found at the blood-brain and brain-cerebrospinal fluid (CSF) barriers in glial cells lining ependyma and pial surfaces in contact with the CSF. Based on this expression pattern, the hypothesis was tested that AQP4 plays a role in the production of brain edema in response to two established neurological insults—acute water intoxication, producing serum hyponatremia and cellular brain edema, and ischemic stroke, producing brain swelling by a combination of cellular and vasogenic edema²⁵. AQP4 deletion conferred remarkable protection from brain edema in these models, with improved mouse survival and clinical outcome, as well as reduced brain swelling. In other recent studies, phenotype differences were found in peritoneal fluid transport²⁶ and dietary fat processing²⁷ in AQP1 null mice, colonic water transport in AQP4 null mice²⁸, and lung water permeability in AQP1 and AQP5 null mice²⁹. However, the tissue-specific expression of an aquaporin does not indicate physiological significance as shown by the unimpaired gastric acid secretion³⁰ and skeletal muscle function³¹ in AQP4 null mice.

AQP1 GENE DELIVERY TO KIDNEY USING ADENOVIRUS

Exploratory studies were done to assess the feasibility of aquaporin gene delivery in kidney. An adenovirus (Ad5-type) encoding AQP1 was generated, purified, and titered. Virus infection of CHO cell cultures gave strong expression of functional AQP1 protein at the cell plasma membrane³². Direct intrarenal injection of the adenovirus gave strong AQP1 protein expression at one week that was limited to the location of the injection site Figure 1a. In preliminary experiments retrograde ureteral virus infusion in rats gave strong AQP1 expression in ureteral and renal papilla with lesser and patchy expression in more cortical regions of collecting duct. Attempts to deliver AQP1 cDNA using liposomal vectors did not give transgene expression by retrograde ureteral and intravenous approaches but was effective by intrarenal injection, but only at the injection site.

Figure 1.

Immunofluorescence localization of AQP1 protein in kidneys of adenovirus-treated AQP1 null mice. (A) AQP1 expression at four days after direct intrarenal injection. *, injection site. (B) Localization of AQP1 in cortex, and (C) medulla at four days after a single intravenous infusion of adenovirus. Scale bar, 100 m.

Full figure and legend (106K)

Based on preliminary investigations, purified adenovirus was administered to AQP1 null mice by tail vein infusion. Some (10–20%) of the mice given adenovirus at the highest doses (10¹⁰ pfu) died within 30 min of the infusion, possibly because of an immune response producing acute hypotension. The remaining mice survived without little apparent morbidity. At 4 days after adenovirus infusion, immunostaining and immunoblot analysis showed strongest AQP1 protein expression in liver, followed by kidney and spleen. AQP1 protein was expressed in hepatic sinusoids, apical and basolateral membranes in renal cortex Figure 1b, and medullary vasa recta Figure 1c. No AQP1 expression was found in glomeruli, limb of Henle, or collecting duct, and no AQP1 was detected in untreated AQP1 null mice. This general pattern of AQP1 protein expression in kidneys of adenovirus-treated mice is consistent with reported studies of adenovirus-reporter and other constructs. By tail-vein infusion, Huard et al.⁸ reported that a luciferase reporter was strongest in liver, followed by diaphragmatic muscle and kidney, and Yayama et al.³³ found strongest kallikrine gene expression in liver, followed by kidney. We note that the precise expression pattern of an introduced gene depends not only on adenovirus dose and accessibility, but on the cell-specific kinetics of transcript and protein synthesis and turnover.

Functional correction of the urinary concentrating defect was evaluated in six AQP1 null mice given a single intravenous infusion of AQP1-Ad5 (10¹⁰ pfu). A 36-hour water deprivation test was done before and at three days after virus infusion. Figure 2a shows little increase in urine osmolality in five out of six mice in the initial water deprivation test, whereas urine osmolality increased substantially in all mice in the second water deprivation test done after the virus treatment. Figure 2b summarizes body weight loss after the first and second water deprivation tests, showing significantly less weight loss after the virus treatment. All of the untreated mice were judged to be lethargic and hypoactive after water deprivation, whereas many of the treated mice were judged to have minimal or no hypoactivity. Figure 2c shows that the functional improvement in urinary osmolality after water deprivation was lost over 3–5 weeks, as is generally found for first-generation adenoviral vectors like the Ad5 used here^7,34,35,36. Viral DNA (by PCR) and AQP1 transcript (by RT-PCR) were detectable by 17 weeks, through levels were substantially reduced compared with 5 weeks³². In a larger group of mice studied at five days after virus infusion, averaged urine osmolality after water deprivation was 838 31 mOsm in the virus-treated AQP1 null mice, significantly greater than that of 676 47 mOsm in the untreated null mice but much less than that in wild-type mice Figure 3a. The incomplete correction is probably related to the lack of AQP1 expression in the thin descending limb of Henle of the virus-treated mice, as well as to the patchy and low-level AQP1 expression in proximal tubule. In 11 mice urine osmolality increased to 900–1220 mOsm, but there was substantial mouse-to-mouse variability, and 25% of the mice showed no significant correction. Figure 3b summarizes the weight loss in the same group of mice after a 36-hour water deprivation. Although all null mice lost substantially more weight than the wild-type mice, there was significantly less weight loss in the virus-treated than the untreated null mice. Water permeability in proximal tubule apical membrane was significantly increased at one week in the virus-treated AQP1 null mice, but barely above baseline by five weeks³².

Figure 2.

Urine osmolality and mouse weight loss in AQP1 null mice before vs. after virus infusion. (A) Urine osmolalities measured at indicated times before and after a 36-hour water deprivation. Water deprivation tests were done prior to and after virus treatment. (B) Body weight loss (expressed as percentage initial weight) in water deprivation tests done before and after virus treatment. *P < 0.05. (C) Increase in urine osmolalities (mean SE) for 36-hour water deprivation tests done at indicated times after virus treatment. Adapted from³².

Full figure and legend (26K)

Figure 3.

Summary of (A) urine osmolalities and (B) weight loss in a series of wild-type, untreated AQP1 knockout, and virus-treated AQP1 knockout mice. Virus-treated knockout mice were given 5-10 10⁹ pfu adenovirus at three or four days prior to water deprivation test. Data (mean SE) are shown for 36-hour water deprivation tests. Numbers of mice in each group are shown in parentheses. * P < 0.05; ** P < 0.01 (Student t test). Adapted from³².

Full figure and legend (19K)

AQUAPORIN GENE DELIVERY TO EXTRARENAL TISSUES

The salivary gland has been proposed as an important target organ for aquaporin gene therapy. As mentioned above, AQP5 null mice manifest defective saliva production²⁷, suggesting that increasing acinar cell water permeability might enhance saliva production in Sjogren's syndrome or post-radiation injury. Several aquaporin gene expression studies were carried out in monolayers of cultured salivary gland cells. Delporte et al.³⁷ infected dog kidney cells and human salivary gland cells with an AQP5 adenovirus and found a significant increase in osmotically driven fluid secretion across cell monolayers. In polarized MDCK cell monolayers, an AQP1 adenovirus increased fluid transport four-fold in response to an osmotic gradient³⁸, and even greater increases in fluid transport were found in virus-infected salivary cell cultures³⁹. Aquaporin gene transfer has also been done in vivo in rat salivary gland. Fluid secretion in irradiated submandibular glands was significantly increased after retrograde ductal instillation of an AQP1 adenovirus, with AQP1 protein expression found in both acinar and ductal epithelial cells⁴⁰. Recently, the safety of a single retrograde AQP1 adenovirus infusion was evaluated in irradiated parotid glands of rhesus monkeys. The gene transfer was well-tolerated; however, there was no consistent increase in saliva secretion⁴¹. It thus remains unclear whether aquaporin gene delivery to salivary gland can be useful in humans.

PROSPECTS FOR AQUAPORIN GENE THERAPY

The initial data using an adenoviral vector supports the feasibility of aquaporin gene replacement in kidney. High-level expression of aquaporins in cells is easily accomplished and is in general not associated with cellular toxicity because aquaporins only facilitate passive water transport driven by osmotic gradients. However, there are substantial problems to overcome in the successful implementation of aquaporin gene replacement in human disease, including gene delivery to targeted cell types and persistent gene expression. Targeting collecting duct epithelial cells in NDI by a retrograde ureteral approach should be feasible, though the noninvasive administration of a targeted construct is preferable. The issue of persistent aquaporin gene expression will require suitable vectors and improved understanding of tubular cell turnover and renal stem cell biology. Transient aquaporin expression in kidney tubules or microvessels might have applications in modifying renal salt and water handling. Finally, outside of the kidney, aquaporin gene delivery may have applications in abnormalities of intracerebral and intraocular pressure regulation, defective exocrine gladular secretions, and other processes where regulation of fluid absorption/secretion or osmolality depends on the magnitude of cell membrane water permeability.

References

References

Acknowledgments

This work was supported by grants DK35124, HL59198, HL60288, and DK43840 from the National Institutes of Health, and Research Development Program grant R613 from the National Cystic Fibrosis Foundation.