TABLE 1


Mean total luciferase in the liver following the intra-
portal injection (over 30 seconds) of 100 μg pBS.CMVLux
in 1 ml of different solutions with no clamp or with the
hepatic vein and inferior vena cava clamped for two minutes.

	Mean
	Luciferase	Standard	Number of
Condition	(total ng/liver)	Error	Livers

no clamp, normal saline solution	0.4	0.7	n = 6
(NSS)
no clamp, 20% mannitol	4.8	8.1	n = 3
clamp, 20% mannitol	14.6	26.3	n = 9
clamp, 2.5 units heparin/ml in	11.8	12.5	n = 4
NSS
clamp, 15% mannitol and 2.5	120.3	101.5	n = 12
units heparin/ml in NSS

Luciferase activities in each liver were evenly distributed in six divided sections assayed (Table 2). All six parts of each liver from all three animals had substantial amounts of luciferase. This is in marked contrast to the direct interstitial, intralobar injection of DNA in which the expression is restricted to the site of injection (R. W. Malone et al., J. Biol. Chem 269, 29903 (1994); M. A. Hickman, et al., Hum. Gene Ther. 5, 1477 (1994) incorporated herein by reference).[0055]

TABLE 2


The distribution of luciferase expression over the six
liver sections in animals injected intraportally (over
30 seconds) with 100 μg of pBS.CMVLux in 1 ml of normal
saline solution plus 15% mannitol and 2.5 units
heparin/ml and with the hepatic vein clamped for 2 minutes.

Total luciferase/Liver (ng/Liver/mouse)

	LiverSection	Mouse #	1	Mouse #2	Mouse #3

½ of median lobe	496.5	66.9	304.5
other ½ of median	177.0	126.1	241.4
lobe
½ of left lateral	763.8	208.7	325.2
lobe
other ½ of left	409.4	160.4	218.9
lateral lobe
right lateral lobe	527.8	129.7	216.2
caudal lobe + small	374.1	149.7	240.8
piece of right lateral
lobe
Total	2,748.6	841.5	1,547.0
Mean	458.1	140.3	257.8
Range	177-763	67-209	216-325
Standard Deviation	194.0	46.6	45.9

Conclusions

1. High levels of luciferase expression were obtained from injecting 100 μg of pBS.CMVLux intraportally.[0056]

2. The highest levels of luciferase expression were obtained when the animals were injected intraportally over 30 seconds with 100 μg of pBS.CMVLux in 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml and with the hepatic vein clamped for 2 minutes.[0057]

3. These high levels of expression were consistently obtained in dozens of mice.[0058]

4. The luciferase expression was evenly distributed throughout the liver.[0059]

Example 2

The effects of other factors on expression were explored using the same methods for the intraportal injection of pBS.CMVLux.[0060]

Methods

Unless otherwise specified, the intraportal injections and luciferase assays were done as in Example 1.[0061]

Results

Compared to the results with 100 μg of pBS.CMVLUX, luciferase expression was not greater with 500 μg of plasmid DNA (Table 3). Luciferase expression was approximately 7-fold less if 20 μg of pBS.CMVLux DNA was injected instead of 100 μg.[0062]

TABLE 3


Total luciferase expression in each liver of each
animal injected intraportally (over 30 sec) with 20 μg,
100 μg, or 500 μg of pBS.CMVLux in 1 ml of normal
saline solution plus 15% mannitol and 2.5 units
heparin/ml and with the hepatic vein occluded for 2 min.

Total luciferase/Liver (ng/Liver/mouse)

	100μg	500 μg
Mouse Number	pBS.CMVLux	pBS.CMVLux

1	1,023	15
2	178	82
3	108	23
4	140	340
Mean	362	115
Standard	441	153
Deviation

The times for which the hepatic vein was occluded were varied from 2 min to 4 min and to 6 min. In Table 4, one can see that the time of occlusion did not have a large effect on expression.[0063]

TABLE 4


Effect of time of hepatic vein occlusion on luciferase
expression in animals injected intraportally with
100 μg of pBS.CMVLux in 1 ml of normal saline
solution plus 15% mannitol and 2.5 units heparin/ml.

Total luciferase/Liver (ng/Liver/mouse)

Mouse Number	2min	4min	6min

1	4.6	1.9	32.7
2	44.9	11.5	6.4

The times over which the injections were done were varied from 30 seconds to 1 minute and 2 minutes. In Table 5, one can see that injecting the 1 ml of the DNA solution (100 μg pBS.CMVLux) over 30 seconds enabled the highest levels of luciferase expression. Longer times of injection led to lower levels.[0064]

TABLE 5


Effect of length of injection (time it took to inject
all of the 1 ml) on luciferase expression in animals
injected intraportally with 100 μg of pBS.CMVLux in 1 ml
of normal saline solution plus 15% mannitol and 2.5 units
heparin/ml and with the hepatic vein occluded for 2 min.

Total luciferase/Liver (ng/Liver/mouse)

	Mouse Number	30sec	1min	2min

1	2,697	188	21.6
2	790	13.4	19.9
3	1,496	141.1	11.8
Mean	1,662	114	18
Standard Deviation	964	91	5

If the total volume of the injection fluid was 0.5 ml instead of 1.0 ml, luciferase expression decreased 70-fold (Table 6) suggesting that 0.5 ml was not sufficient to fill the intravascular space and distribute the DNA throughout the parenchyma.[0065]

TABLE 6


Total luciferase expression in each liver of each
animal injected intraportally (over 30 sec) with
100 μg of pBS.CMVLux in either 0.5 or 1 ml of normal
saline solution plus 15% mannitol and 2.5 units
heparin/ml and with the hepatic vein occluded for 2 min.

	Total luciferase/Liver
	(ng/Liver/mouse)

Mouse Number	0.5ml	1ml

1	1.6	51.9
2	4.7	124.8
3	0.4	266.9
Mean	2.3	147.9
Standard	2.3	109.4
Deviation

Conclusions

1. The optimal conditions are in fact the conditions first described in example 1: the animals were injected intraportally over 30 seconds with 100 μg of pBS.CMVLux in 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml and with the hepatic vein clamped for 2 minutes.[0066]

2. Use of 500 μg of pBS.CMVLux did not enable greater levels of expression but expression was approximately7-fold less if 20 μg of DNA was used.[0067]

3. Occluding the hepatic vein for longer than 2 minutes did not increase expression.[0068]

4. Injecting the pBS.CMVLux over 30 seconds gave the highest luciferase levels as compared to injection times longer than 30 seconds.[0069]

5. Injecting the pBS.CMVLux in 1 ml gave higher luciferase levels than injecting the pBS.CMVLux in 0.5 ml.[0070]

Example 3 Methods

The intraportal injections and luciferase assays were performed as in Example 1 except that some animals received daily subcutaneous injections of 1 mg/kg of dexamethanone (Elkins-Sinn, Cherry Hill, N.J.) starting one day prior to surgery. The conditions for the injections were intraportal injections over 30 seconds with 100 μg of pBS.CMVLux in 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml and with the hepatic vein clamped for 2 minutes.[0071]

Results

Under the conditions described above (i.e., hypertonic solution containing heparin and hepatic vein closure) into animals that had been injected with daily injections of dexamethasone starting the day prior to plasmid injection, luciferase expression was three-fold greater than the expression without dexamethasone (Table 7).[0072]

TABLE 7


The effect of dexamethasone injections on luciferase
expression after the intraportal injection of pBS.CMVLux.

Total luciferase/Liver (ng/Liver/mouse)

Mouse Number	NO Dexamethasone	WITHDexamethasone

1	51.9	1,181.1
2	124.8	364.7
3	266.9	82.8
4	73.7	120.5
5	52.6	1,022.9
6	7.3	178.1
7	146.1	107.6
8	231.4	140.2
9	271.2
10	8.7
11	8.3
12	201.1
Mean	120.3	399.8
Standard	101.4	444.1
Deviation

Dexamethasone could have increased the production of luciferase and the expression of other genes by several mechanisms. They include increasing the amount of plasmid DNA that enters the liver cells by modifying the state of the liver cells. It could also help the liver cells withstand the high pressure. However, the most likely mechanism is that dexamethasone directly stimulates the CMV promoter and thereby directly increases expression of luciferase by stimulating transcription of the luciferase messenger RNA.[0073]

The use of dexamethasone demonstrates that using a readily available pharmaceutical, the levels of expression can be substantially increased and regulated.[0074]

Conclusion

1. Dexamethasone administration increased luciferase expression from intraportally-injected pBS.CMVLux plasmid DNA three-fold.[0075]

2. This demonstrates that the expression from the liver can be regulated using a commonly-used pharmaceutical.[0076]

Example 4 Methods

The intraportal injections were performed using the previously stated technique of injections over 30 seconds with 100 μg of plasmid DNA in 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml and with the hepatic vein clamped for 2 minutes. The mice also received daily subcutaneous injections of 1 mg/kg of dexamethasone (Elkins-Sinn, Cherry Hill, N.J.) starting one day prior to surgery.[0077]

The plasmids pBS.CMVLacZ and pBS.CMVnLacZ were used to express a cytoplasmic and nuclear β-galactosidase protein, respectively, from the CMV promoter (Picard, D. & Yamamoto, K. EMBO J. 6:3333-3340, 1987; incorporated herein by reference). They were constructed by placing either a 3.5-kg-HindIll/Xbal B-galactosidase sequence from pSDKLacZpa (Danko, I. et al. Gene Therapy 1:114-121, 1994; incorporated herein by reference) or a sequence encoding a nuclear-localizing-galactosidase (Picard, D. & Yamamoto, K. EMBO J. 6:3333-3340, 1987; incorporated herein by reference) into pBlueCMV (Danko, I. et al. Gene Therapy 1:114-121, 1994; incorporated herein by reference).[0078]

Two days after intraportal injection, the livers were perfused with 1% paraformaldehyde and 1.25% glutaraldehyde in phosphate buffered saline (PBS) and then kept in this solution for one day. After the livers were stored in 30% sucrose, they were cryosectioned. The sections were mounted on slides and stained for 1 hour to one day with a PBS solution (pH 7.5) containing 400 μg/ml X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactoside) (Sigma), 5 mM potassium ferricyanide, 5 mM ferrocyanide, and 1 mM MgCl[0079]₂. After washing, the sections were then counter-stained with hematoxylin and eosin. In the livers injected with the nuclear-localizing β-galactosidase vector, the washing step after hematoxylin incubation was omitted to decrease its nuclear staining.

Results

Having defined the optimal conditions, the types and percentages of transfected cells were determined. After injections of a 100 μg of the cytoplasmic (pBS.CMVLacZ) or the nuclear (pBS.CMVnLacZ) β-galactosidase expression vectors into dexamethasone-treated animals, liver cryosections 10- to 30-μm thick were stained for β-galactosidase using X-gal at pH 7.5 to prevent background staining. Intense blue staining was observed in approximately 1% of the liver cells and was evenly distributed throughout the liver. X-gal incubations for only 1 hour resulted in intensely blue cells; suggesting that the transfected cells expressed relatively large amounts of the foreign genes. Control livers injected with 100 μg of pBS.CMVLux did not contain any positively-stained cells. Necrosis was observed in approximately 10% of the sections. However, some livers with high β-galactosidase expression did not contain any sections with necrosis.[0080]

The hepatocytes were identified by their characteristic morphology. For example, many of the cells in the livers injected with the nuclear β-galactosidase vector, pBS.CMVnLacZ, had blue staining in two nuclei, which is a trait only of hepatocytes. Although the majority of the positively-stained cells were hepatocytes a few small, non-hepatocyte cells contained blue staining.[0081]

Conclusion

1. Approximately 1% of the liver cells were transfected with the B-galactosidase gene throughout the entire liver.[0082]

2. Almost all of the transfected liver cells were hepatocytes.[0083]

Example 5 Methods

Luciferase expression in the liver was compared to that in cultured HepG2 hepatocytes in 35-mm plates. Transfections were done using 3 μg of pBS.CMVLux/plate and either 3 μg of Lipofectin (Life Technologies, Bethesda, Md.) or 6 μg of LipofectAMINE (Life Technologies, Bethesda, Md.) per manufacturer's instructions. Two days after transfection, 200 ul of lysis buffer was added to the cultures and 20 ul of the supernatant were analyzed for luciferase activity as in Example 1.[0084]

Results

The efficiency of luciferase expression using this technique was compared to other methods of gene transfer both in vitro and in vivo. Transfections performed under optimal conditions with pBS.CMVLUX and Lipofectin or LipofectAMINE (Life Technologies Inc.) in HepG2 hepatocytes in culture (n=8) yielded a mean total of 3.7±4.5 ng luciferase/35-mm plate and 2.8±2.0 ng luciferase/35-mm plate. Thus the efficiency of transfection (without dexamethasone) in terms of ng of luciferase/μg of pBS.CMVLUX DNA was approximately 1 ng/μg both in vitro and in vivo.[0085]

The published procedure of repetitively and directly injecting naked plasmid DNA into a rat liver lobe was reduced proportionately for mouse liver (R. W. Malone et al., J. Biol. Chem 269, 29903 (1994); M. A. Hickman, et al., Hum. Gene Ther. 5, 1477 (1994); incorporated herein by reference). A total of 100 μg of pBS.CMVLUX in a total volume of 200 ul of normal saline was injected within five different sites (40 ul/site) into the left lateral lobe of 30 g mice treated with dexamethasone. A mean total of only 0.1 ng/liver (4 livers; 0.001 ng luciferase/μg DNA) was obtained and the luciferase expression was only present in the injected lobe. Approximately 30-fold more luciferase expression was obtained if the direct intralobar injections were done using 1 ml of injection fluid and clamping the hepatic vein. In the previous studies involving the multiple injections of a total of 500 μg of pCMVL into a liver lobe of dexamethasone-treated rats, a mean of 9.87 ng of luciferase/liver (0.02 ng/μg DNA) was expressed (R. W. Malone et al.,[0086]J. Biol. Chem269, 29903 (1994); M. A. Hickman, et al.,Hum. Gene Ther.5, 1477 (1994)).

With regard to muscle, we typically inject 10 μg of pBS.CMVLUX or pBS.RSVLUX ((Danko, I. et al. Gene Therapy 1:114-121, 1994)) in normal saline into 6-8 mouse quadriceps muscle per experiment. In dozens of experiments, mean total luciferase per muscle was 0.4-1 ng (±0.5-1.2) and the efficiency was 0.04-0.1 ng luciferase/μg DNA.[0087]

TABLE 8


Comparison of efficiency of gene transfer in
terms of luciferase expressed per μg of pBS.CMVLux
plasmid DNA used for the method (ng luciferase/μg DNA).

	Mean Total Yield	Amount of	Efficiency
Method of	Of	pBS.CMVLux	(ng Luciferase/
Gene Transfer	Luciferase (ng)	Used (μg)	μg DNA)

Intraportal Mouse	120.3 ± 101.5	100	1.2
Liver (above	n = 12
optimal conditions-
Table 1)
hepatic vein
clamped
HepG2 In Vitro	3.7 ± 4.5	3	1.2
with Lipofectin	(n = 8)
HepG2 In Vitro	2.8 ± 2.0	3	0.9
with	(n = 8)
LipofectAMINE
Intralobar Mouse	0.1 ± 0.1	100	0.001
Liver (20 ul/site × 5	n = 4
sites) hepatic vein
not clamped
Intralobar Mouse	2.8 ± 5.6	100	0.028
Liver (1 ml/1 site)	n = 4
hepatic vein
clamped
Intralobar Rat Liver	9.87	500	0.02
(5 sites)
from published data
(Journal of Biologic
Chemistry
269:29903, 1994;
Human Gene
Therapy 5:1477,
1994)
Intramuscular	0.4-1 ± 0.5-1.2	10	0.04-0.1
	n > 50

Conclusions

1. The intraportal delivery of naked DNA was more than an order of magnitude more efficient than interstitial delivery into either liver or muscle and more evenly distributed.[0088]

Example 6 Methods

The intraportal injections were done using the above optimal injections which are intraportal injections over 30 seconds with 100 μg of pCMVGH in 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml and with the hepatic vein clamped for 2 minutes. Some animals received daily intramuscular injections of 100 mg/kg of cyclosporine (Sandimmune, Sandoz) or daily subcutaneous injections of 1 mg/kg of dexamethasone (Elkins-Sinn, Cherry Hill, N.J.), or both starting one day prior to surgery.[0089]

The previously described pCMVGH plasmid DNA was used to express human growth hormone (hGH) (C. Andree, et al., Proc. Natl. Acad. Sci. U.S.A. 91, 12188 (1994); incorporated herein by reference). Blood obtained from the retro-orbit sinus was analyzed for serum concentration of hGH using the radioimmune assay (RIA), HGH-TGES 100T kit from Nichols Institute (San Juan Capistrano, Calif.).[0090]

Results

Human growth hormone (hGH) was used as a marker gene to assess the ability of this gene transfer technique to produce a therapeutic serum protein (Table 9). Two days after the intraportal injection of 100 μg of pCMVGH under the above optimal conditions, the mean hGH serum concentration was 57±22 ng/ml (n=12) with a range of 21-95 ng/ml. Neither dexamethasone nor cyclosporine pre-treatment significantly affected these initial hGH levels. In two animals injected with pBS.CMVLUX, background hGH levels were 0.3±0.1 ng/ml for 4 weeks afterwards.[0091]

In humans, normal pulsatile levels of GH peak at approximately 20 ng/ml above baseline values of approximately 1 ng/ml and can attain concentrations of 10-180 ng/ml after growth hormone releasing hormone (GHRH) stimulation (A. Favia, J. D. Veldhuis, M. O. Thomer, M. L. Vance, J. Clin. Endocrinol. Metab. 68, 535 (1989); R. W. Holl, M. L. Hartman, J. D. Veldhuis, W. M. Taylor, M. O. Thorner, J. Clin. Endocrinol. Metab. 72, 854 (1991); W. S. Evans et al., Am. J. Physiol. 252, E549 (1987); F. P. Alford, H. W. G. Baker, H. G. Burger, J. Clin Endocrinol. Metab. 37, 515(1973); incorporated herein by reference). The half-life of hGH is approximately 20 min in humans and 4.5 min in mice; hence these serum levels could translate into much higher levels for more stable proteins (S. Peeters and H. G. Friesen, Endocrinol.101, 1164 (1977); A. Favia, J. D. Veldhuis, M. O. Thomer, M. L. Vance, J. Clin. Endocrinol. Metab. 68, 535 (1989); incorporated herein by reference). For example, if a protein such as alpha-antitrypsin has a half-life that is ten times longer than human GH, then the circulating blood levels should be at more than ten times higher given the same efficiency of protein production. Another example is that for hemophilia which requires levels of factor VIII or IX in the range of approximately 1 μg of the clotting factor/ml of blood. Given the increased stability of these clotting factors, then the 0.1 μg/ml of hGH that we can achieve after intraportal injection of the respective gene means that we would be able to obtain therapeutic levels of clotting factors to prevent bleeding in patients with hemophilia. In summary, these results demonstrate that the intraportal naked DNA technique could be used to produce therapeutic levels of a circulating blood protein.[0092]

Serial measurements of hGH serum levels enabled the stability of expression in individual mice to be assessed (Table 9). In untreated animals, hGH expression was unstable as in previous studies in which the plasmid DNA was delivered to non-hepatectomized livers using polylysine complexes or intralobar injections of naked DNA.[0093]

An immune response could kill hepatocytes expressing the human protein. To test the hypothesis that expression was unstable because of an immune response, hGH levels were followed in animals that received cyclosporine with or without dexamethasone administration (Table 9). After an acute drop off, hGH levels remained at 6-11 ng/ml for four weeks in animals that received both dexamethasone and cylcosporine. In animals that received dexamethasone alone or cyclosporine alone, hGH expression was prolonged as compared to the non-treated animals but not to the same extent as the animals that received both agents. The ability for this gene transfer method to enable expression of a foreign gene should increase its utility.[0094]

TABLE 9


Mean serum levels (ng/ml of serum) of human growth hormone
(hGH) following intraportal administration of pCMVGH under
optimal conditions in mice (2 to 3 animals for each timepoint)
receiving various treatments. Optimal conditions are defined as the
use of 0.9% saline, 15% mannitol, 2.5 units/ml heparin solution
that was intraportally injected with the hepatic vein closed.

DAYS
AFTER
INJECTION	NONE	CSA alone	DEX alone	CSA +DEX

2	69	43	72	51
4	11	8	14	14
8	3	6	7	13
12	0	4	7	15
15	0	3	5	13
21	0	1.5	2.6	9.7
28	0	1	2.2	7.9

Conclusions

1. These results demonstrate that the intraportal naked DNA technique could be used to produce therapeutic levels of a circulating blood protein that is currently used to treat humans.[0095]

2. The levels of the circulating blood protein (i.e. hGH) remained elevated for at least one month after a single injection.[0096]

Example 7 Methods

After the portal veins of 25 g, 6-week old mice were exposed through a ventral midline incision, 100 μg of pBS.CMVLux plasmid DNA in 0.5 ml or 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml were manually injected over 30 seconds into the portal vein near the junction of the splenic vein and portal vein. The portal vein had two clamps placed distal and proximal to the point of injection so as to direct the injection fluid into only the splenic vein and to prevent the injection fluid from going to the liver or intestines. The injections were done using a 30-gauge, ½-inch needle and 1-ml syringe. 5×1 mm, Kleinert-Kutz microvessel clips (Edward Weck, Inc., Research Triangle Park, N.C.) were used. Anesthesia was obtained from intramuscular injections of 1000 μg of ketamine-HCl (Parke-Davis, Morris Plains, N.J.) and methoxyflurane (Pitman-Moore, Mudelein, Ill. USA) which was administered by inhalation as needed. was purchased from Sigma. Heparin was purchased from LyphoMed (Chicago, Ill.).[0097]

Two days after injection the spleens and pancreas were removed and placed in 500 ul of lysis buffer and 20 ul were analyzed for luciferase expression as described above.[0098]

Results

Substantial amounts of luciferase activity were obtained in the spleen and pancreas of all four mice with both injection fluids of 0.5 ml and 1 ml.[0099]

TABLE 10


Luciferase expression after the intravascular-administration
of pBS.CMVLux into the splenic vein via the portal vein.

	Total luciferase/Organ
	(pg/organ/mouse)

Injection Volume	Spleen	Pancreas

0.5 ml	814.4	97.2
0.5 ml	237.3	88.7
1 ml	168.7	109.4
1 ml	395.0	97.7
Mean	403.9	98.3
Standard	289.6	8.5
Deviation

Conclusion obtained from intramuscular injections of 1000 μg of ketamine-HCl (Parke-Davis, Morris Plains, N.J.) and methoxyflurane (Pitman-Moore, Mudelein, Ill. USA) which was administered by inhalation as needed. was purchased from Sigma. Heparin was purchased from LyphoMed (Chicago, Ill.).

Two days after injection the spleens and pancreas were removed and placed in 500 ul of lysis buffer and 20 ul were analyzed for luciferase expression as described above.[0100]

Results

Substantial amounts of luciferase activity were obtained in the spleen and pancreas of all four mice with both injection fluids of 0.5 ml and 1 ml.[0101]

TABLE 10


Luciferase expression after the intravascular-administration of
pBS.CMVLux into the splenic vein via the portal vein.
Total luciferase/Organ (pg/organ/mouse)

Injection Volume	Spleen	Pancreas

Conclusion

1. Intravascularly-administered plasmid DNA can express efficiently in spleen and pancreas.[0102]

Example 8 Methods

100 μg of pBS.CMVLux in 10 ml of normal saline solution plus 15% mannitol was injected into the femoral artery of adult rats with the femoral vein clamped. One to four days after injection, the quadricep was removed and cut into 10 equal sections. Each sections were placed into 500 ul of lysis buffer and 20 ul were assayed for luciferase activity as described above.[0103]

Results

Substantial amounts of luciferase expression were expressed in the quadriceps following the intravascular delivery of plasmid DNA.[0104]

TABLE 11


Luciferase expression in the quadricep of a rat after the injection of
100 μg of pBS.CMVLux into the femoral artery and with the femoral
vein clamped.

		Total Luciferase
	Rat Number	(pg/quadriceps)

	1	157.5
	2	108.8
	3	139.2
	4	111.3
	Mean	129.2
	Standard Deviation	23.4

Conclusion

1. Intravascularly-administered plasmid DNA can express efficiently in muscle.[0105]

Example 9

The previous examples involved injections into the afferent blood vessels of organs. In the liver, the hepatic vein is an efferent blood vessel since it normally carries blood away from the liver into the inferior vena cava. Also in the liver, the portal vein and hepatic arteries are afferent blood vessels in relation to the liver since they normally carry blood towards the liver.[0106]

These set of experiments were designed to determine whether plasmid DNA could be efficiently expressed if delivered by a retrograde route into the efferent vessel of the liver (i.e. the hepatic vein).[0107]

Since another luciferase expression vector was used, pCILuc, the results obtained with the hepatic vein injections were directly compared to results using the above technique of injecting the portal vein.[0108]

Methods

100 μg of pCILuc in 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml were injected over 30 seconds into hepatic vein via the inferior vena cava. Since it was difficult to directly inject the hepatic vein in rodents, the injections were directed into the inferior cava which was clamped in two locations; proximal and distal (i.e. downstream and upstream) to the entry of the hepatic vein into the inferior vena cava. Specifically, the downstream inferior vena cava clamp was placed between the diaphragm and the entry point of the hepatic vein. The upstream inferior vena cava clamp was placed just downstream of the entry point of the renal veins. Therefore, the 1 ml of the injection fluid entered the hepatic vein and the liver. Since the veins of other organs such as the renal veins enter the inferior vena cava at this location, not all of the 1 ml of injection fluid goes into the liver.[0109]

In some of the animals that received retrograde injections in the inferior vena cava, the hepatic artery, mesenteric artery, and portal vein were clamped (occluded) for approximately five minutes immediately before and then after the injections. Specifically, the order of placing the clamps were as follows: first on hepatic artery, then portal vein, then downstream vena cava, and then upstream vena cava. It took about three minutes to place all these clamps and then the injections were done. The clamps were left in place for an additional two minutes from the time that the last clamp (upstream vena cava clamp) was placed.[0110]

The intraportal injections were performed as stated using optimal intraportal injections over 30 seconds with 100 μg of pCILuc in 1 ml of normal saline solution plus 15% mannitol and 2.5 units heparin/ml and with the hepatic vein clamped for 2 minutes.[0111]

Some of the mice also received daily subcutaneous injections of 1 mg/kg of dexamethasone (Elkins-Sinn, Cherry Hill, N.J.) starting one day prior to surgery.[0112]

The pCILuc plasmid expresses a cytoplasmic luciferase from the CMV promoter. It was constructed by inserting the cytoplasmic luciferase cDNA into the pCI (Promega Corp., Madison, Wisc.) CMV expression vector. Specifically, a Nhel/EcoRI restriction digestion fragment containing the cytoplasmic luciferase cDNA was obtained from pSPLuc (Promega Corp.) and inserted into pCI plasmid DNA that was digested with NheI and EcoRI, using conventional recombinant DNA techniques (Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) in Molecular Cloning Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.).[0113]

Two days after the injections, the luciferase activity was measured as above in six liver sections composed of two pieces of median lobe, two pieces of left lateral lobe, the right lateral lobe, and the caudal lobe plus a small piece of right lateral lobe.[0114]

Results

A. Inferrior Vena Cava/Hepatic Vein Injections with the Portal Vein and Hepatic Arterty Clamped (*Injections in[0115]animal #3 were not optimal since the fluid leaked during the injections.) Injections were done in 6-week old animals that received dexamethasone.



	Luciferase Activity (ng)

Sections	Animal #	1	Animal #2	Animal #3*

1	5,576.7	4,326.4	1,527.4
2	8,511.4	4,604.2	1,531.6
3	5,991.3	5,566.1	2,121.5
4	6,530.4	9,349.8	1,806.3
5	8,977.2	4,260.1	484.2
6	9,668.6	6,100.2	1,139.3
total liver	45,255.5	34,206.9	8,610.4
mean	29,357.6
standard deviation	18,797.7

B. Vena Cava/Hepatic Vein Injections with the Portal Vein and Hepatic Artery not Clamped. Injections were done in 6-week old animals that did not receive dexamethasone.[0116]



	Luciferase Activity (ng)

Sections	Animal #	1	Animal #2

1	360.6	506.2
2	413.5	724.7
3	463.0	626.0
4	515.5	758.6
5	351.6	664.8
6	437.8	749.6
total liver	2,542.0	4,029.8
mean	3,285.9
standard deviation	1,052.1

C. Portal Vein Injections with the Hepatic Vein Clamped in 6 month old mice that received dexamethasone.[0117]



	Luciferase Activity (ng)

Sections	Animal #	1	Animal #2	Animal #3

1	287.4	417.0	129.2
2	633.7	808.1	220.5
3	689.8	1,096.5	328.2
4	957.8	1,056.9	181.6
5	660.7	1,487.4	178.6
6	812.4	1,276.4	233.4
total liver	4,041.8	6,142.2	1,271.5
mean	3,818.5
standard deviation	2,443.0

D. Portal Vein Injections with the Hepatic Vein Clamped in 6 week old mice that received dexamethasone.[0118]



	Luciferase Activity (ng)

Sections	Animal #	1	Animal #2	Animal #3

1	352.9	379.1	87.0
2	667.5	373.9	108.2
3	424.8	1,277.9	178.4
4	496.3	1,308.6	111.9
5	375.2	296.4	162.3
6	434.7	628.7	123.0
total liver	2,751.4	4,264.7	770.9
mean	2,595.7
standard deviation	1,752.1

E. Summary Table Comparing the Luciferase Expression Obtained Using the Above Conditions[0119]



	Mean Total Luciferase/Liver	Times
Injection Condition@	(μg/liver)	Condition D.

Condition A	29.4	11.3 X
Condition B	3.3	1.27 X
Condition C	3.8	1.46 X
Condition D	2.6	1.00 X

Conclusions

1. Retrograde delivery of plasmid DNA into the efferent vessels of the liver via the hepatic vein/inferior vena cava leads to high levels of gene expression.[0120]

2. The highest levels were achieved using this retrograde approach if the afferent vessels to the liver (portal vein and hepatic artery) were occluded.[0121]

3. The CILuc plasmid enabled much higher levels of luciferase expression than the pBS.CMVLux plasmid (see above examples) using the portal vein approach in both 6-week old and 6-month old mice.[0122]

4. Under all conditions, luciferase expression was evenly distributed throughout all six liver sections.[0123]

Example 10

Animals that received injections into the inferior vena cava were assayed for luciferase to determine whether retrograde delivery into the efferent vessels (veins) of other organs enable gene expression.[0124]

Methods

In the same animals that were injected using condition A above (Inferior Vena Cava/Hepatic Vein Injections with the Portal Vein and Hepatic Artery Clamped in 6 week-old animals that received dexamethasone), the kidneys were removed and assayed for luciferase as described above.[0125]

In the same animals that were injected under condition B above (Inferior Vena Cava/Hepatic Vein Injections with the Portal Vein and Hepatic Artery NOT Clamped in 6-week old animals that did not receive dexamethasone), the adrenal gland and diaphragm muscle, abdominal muscles, and back muscles were removed for luciferase analysis.[0126]

Results

A. Luciferase Activity in Kidneys in Animals Injected Under Condition A.[0127]



	Total Luciferase Activity/Kidney (pg/kidney)

	Animal #1	Animal #2	Animal #3*

Right Kidney	10,827.8	7,662.3	636.3
Left Kidney	733.1	753.8	479.7

B. Luciferase Activity in Adrenals and Various Muscles Injected Under Condition B.[0128]



	Total Luciferase Activity/Tissue (pg/tissue)

	Animal #7	Animal #8	Animal #9

right adrenal	not assayed	82.0	49.9
left adrenal	not assayed	48.4	42.2
diaphragm	41.9	67.9	117.6
abdomen	40.4	43.9	44.0
back	37.7	40.1	40.9

Conclusions

1. Retrograde delivery of plasmid DNA into the efferent vessels of several different tissues led to substantial levels of foreign gene expression in the tissues.[0129]

2. These tissues include the adrenal glands (suprarenal glands), the diaphragm muscle, back muscles and abdominal muscles.[0130]

3. Foreign gene expression in the diaphragm would be especially useful for Duchennes muscular dystrophy since humans with this disorder die from respiratory failure due to fibrosis of the diaphragm muscle[0131]

Example 11

This example explores the use of more accessible vessels such as the hepatic vein and the bile duct, for delivering the naked pDNA in mice. Efficient gene expression was obtained using these efferent delivery routes. Occlusion of other vessels to restrict outflow of the injection solutions enhanced but was not critical for efficient expression. Repetitive injections into the bile duct were also accomplished. Preliminary results are also presented in larger animals, the rat and dog. The incorporation of these findings into laboratory and clinical protocols is discussed.[0132]

Materials and Methods

Plasmid Constructs: The pCILuc plasmid expresses a cytoplasmic luciferase from the human CMV immediately early (hCMV ID) promoter. It was constructed by inserting the luc+ gene, an Nhe I- EcoR I luc+ fragment from pSLuc+ (Promega, Madison, Wisc.), into the pCI expression vector (Promega). pCILux expresses peroxisomal luciferase under control of the hCMV IE promoter. It was constructed by inserting the luciferase gene (Hind III - Bam HI fragment from pBlueCMVLux) into the Sma I site of pCI. pCILacZ was constructed by placing the E. coli LacZ gene (Pst I - Apa I fragment pBS-RSV-LacZ) into the pCI vector (Sma I site). The pCMVGH construct was previously described (Andree, et al., 1994).[0133]

Injection Methods: Plasmid delivery into the hepatic vessels was performed in 6-week old ICR mice, 2.5-6.25 month-old, 200-300 gram Sprague-Dawley rats, and beagle dogs. Ventral midline incisions were performed to expose the liver and the associated vessels. The mice were anesthetized with intramuscular injections of 1000 μg of ketamine-HCl (Parke-Davis, Morris Plains, N.J.) and methoxyflurane (Pitman-Moore, Mudelein, Ill.) which was administered by inhalation as needed. The rats were anesthetized with ether and the dogs were anesthetized with halothane by inhalation. The pDNA was injected in solutions containing 2.5 units/ml or heparin (Lypho-Med, Inc., Chicago, Ill.) (Qian et al., 1991) and either normal saline (0.9% NaCl) or 15% mannitol in normal saline (Sigma Chemical Co., St. Louis, Mo.). All animals received humane care in compliance with institutional (IACUC) guidelines.[0134]

In mice, the intraportal injections were performed as previously described (Budker, et al., 1996). 100 μg of pDNA in 1 ml were manually injected over ˜30 see using a 30-gauge, ½-inch needle and 1-ml syringe without occluding the portal vein upstream from the point of injection. In some animals, a 5×1 mm, Kleinert-Kutz microvessel clip (Edward Week, Inc., Research Triangle Park, N.C.) was applied during the injection at the junction of the hepatic vein and caudal vent cava.[0135]

DNA was delivered in mice to the hepatic vein via an occluded IVC. Clamps (6×1 mm, Kleinert-Kutz curved microvessel clip (Edward Week, Inc., Research Triangle Park, N.C.) were applied downstream (toward the heart) of the hepatic vein and upstream (towards the legs) of the hepatic and renal veins. Injections were done upstream of the hepatic vein. In some of the injections, the portal vein and hepatic artery were clamped using 6×1 mm, Kleinert-Kutz curved microvessel clips. The IVC mouse injections were also performed with 100 μg of pDNA in 1 ml that were manually injected over -30 sec with a 30-gauge, ½-inch needle and 1-ml syringe.[0136]

The bile duct injections in mice were performed using manual injections with a 30-gauge, ½-inch needle and 1-ml syringe. A 5×1 mm, Kleinert-Kutz microvessel clip was used to occlude the bile duct downstream from the point of injection in order to prevent flow to the duodenum and away from the liver. The gallbladder inlet was not occluded. In some of the bile duct injections, the junction of the hepatic vein and caudal vena cava clamped as above. In yet other injections, the portal vein and hepatic artery were clamped in addition to the occlusion of the hepatic vein.[0137]

In mice, repetitive injections into the bile duct were done by placing a polyethylene tube (I.D. 0.28 mm, O.D. 0.61 mm; Intramedic Clay Adams Brand, Becton Dickinson Co., Sparks, Md., USA) catheter into the bile duct after making a hole with a 27 gauge needle. The tubing was secured by a suture around the bile duct and tubing; thereby occluding the bite duct. The other end of the tubing was placed outside the skin of the animal's back so that surgery was not required for repeat injections. No blood vessel occlusions were done for these repetitive administrations. After completion of the studies, anatomical examination indicated that the catheter remained in the bile duct.[0138]

In rats, the intraportal, IVC, and bile duct injections were done as in mice but with the following modifications. The injections were done through a 25-g butterfly needle using a peristaltic pump (Preston varistaltic power pump, Manostat Corp., New York, N.Y.) over 1 or 3 minutes. The downstream IVC clamps in the IVC injections were done downstream of the kidneys. For the portal vein injections, the portal vein and hepatic artery were clamped. The outflow through the hepatic vein was restricted in some animals by clamping the upstream and downstream IVC. In some animals the livers were first flushed with normal saline prior to DNA injection. Rat bile duct injections were done the same as mice. The rat doesnt have a gallbladder.[0139]

In some of the rat portal vein injections, a 25 G needle connected to a pressure gauge (Gilson Medical Electronics, Model ICT-1 1 Unigraph), was inserted into the liver parenchyma to determine the peak pressure within the liver during the injections. The statistical relationship between pressure and luciferase was done using Spearman's rank correlation. A smoothing spline regression model describing the relationship between log luciferase and pressure was estimated using generalized additive model methodology (Hastie, 1992). Akaike's information criteria (Akaike,. 1973) indicated that a smoothing spline with 2 degrees of freedom resulted in the best fit over all integer degrees of freedom between 0 and 5.[0140]

The injections in dogs were done as in rats except that an 18 gauge, 2 inch angiocath (Becton Dickinson, San Jose, Calif.) was used. All dogs except[0141]dog #1 were females. Table 1 indicates the injection conditions. For the bile duct injections, a suture was applied to occlude transiently the bile duct downstream from the point of injection. A DeBakey multipurpose vascular clamp was applied to the cystic duct during injection to prevent the injectate from entering the gallbladder. In dogs, the DNA was pCILux.

Protein Assays: The luciferase assays were done as previously reported (Wolff, et al., 1990). One day after pCILuc injections, the animals were sacrificed and the rodent livers were divided into 6 sections composed of right lateral lobe, caudate lobe, two pieces of median lobe and two pieces of left lateral lobe. For each of the six pieces, 0.7 ml of lysis buffer (0.1% Triton X-100, 0.1 M potassium phosphate, 1 mM DTF pH 7.8) was used for mice and four ml of lysis buffer were used for rat liver. For the dog livers, approximately 10% of each lobe was divided into five to 20 pieces and placed into two ml of lysis buffer. The samples were homogenized using a[0142]PRO 200 homogenizer (PRO Scientific Inc., Monroe, Conn.) and centrifuged at 4,000 rpm for 10 min. at 4° C. Twenty μl of the supernatant was analyzed for luciferase activity. Relative light units (RLU) were converted to pg of luciferase using standards from Analytic Luminescence Laboratories (ALL, San Diego, Calif.). Luciferase protein (pg)=5.1×10⁻⁵×RLU+3.683 (r²=0.992).

Ten-μm thick tissue sections were stained for β-galactosidase expression as previously described using 1-4 hour X-GAL incubations (Budker, et al., 1996). Hematoxylin was used for the counterstain but the alkaline step was omitted so that the hematoxylin stain remained red. The percent of blue-stained cells in the liver sections was determined by counting 3000 cells in three sections and averaging.[0143]

Blood obtained from the retro-orbit sinus was analyzed for serum concentration of hGH using the RIA, HGH-TGES 100T kit from Nichols Institute (San Juan Capistrano, Calif.). Serum ALT and GGT levels were done using EKTACHEM DT slides and a[0144]KODAK EKTACHEM DT 60 ANALYZER as recommended by the manufacturer (Kodak, Rochester, N.Y.).

Mice Luciferase Experiments: Our previous studies used the pBS.CMVLux plasmid for evaluating the optimal conditions for naked pDNA expression following importal injection. These optimal conditions were intraportal injections of 100 μg of pDNA in 1 ml of 15% mannitol and 2.5 units heparin/ml in normal saline solution. The injections were done over 30 seconds with the hepatic vein and IVC occluded. In this study, 100 μg of pCILuc injected under similar conditions yielded a mean total luciferase protein/liver of 3.73 μg/liver (FIG. 1A, PV+CL), approximately 30- times greater than that obtained with pBS.CMVLux. Part of this increase could be attributed to greater operator experience with these injection techniques. Injections with pCILuc under these conditions without clamping the hepatic vein yielded approximately 750-fold less luciferase (FIG. 1A, PV−CL).[0145]

The hepatic veins (via the IVC) of another set of mice were injected with 100 μg of pCILuc in 1 ml of 15% mannitol and 2.5 units heparin/mil in normal saline solution. A mean total luciferase protein/liver of 17.34 μg/liver (FIG. 1A, IVC+CL) was obtained when the portal vein was clamped as compared to a mean total luciferase protein/liver of 2.83 μg/liver (FIG. 1A, IVC−CL) without occluding the portal vein.[0146]

Serum ALT and GGT assays were performed on mice one and eight days after each of the above injections with pCILuc (4 mice for each condition). No increases in GOT were observed after any of the injections including the bile duct injections. Serum ALT levels increased to 200-400 U/L one day after portal vein and bile duct injections. One day after IVC injections serum ALT levels increased to ˜1500 U/L in half of the mice but was only ˜250 U/L in the other half. By eight days after injection, serum ALT levels decreased to baseline levels in all animals. For positive control purposes, a non-lethal intraperitoneal injection of 40 μl of 50%, carbon tetrachloride in mineral oil was performed. An average of 25,900 U/L (n=4) was observed one day after injection.[0148]

Rat and Dog Luciferase Experiments: Similar injections into the portal vein, IVC (to the hepatic vein), and bile duct were done in rats (FIG. 1B). For the portal vein injections, the injection volumes were increased 15-times over that used in mice because the rat livers are 15-times larger than mouse livers. One day after 750 of μg of pCILuc in 15 ml of 15% mannitol and 2.5 units heparin/mi in normal saline solution were injected into the portal vein while occluding the hepatic vein, an average of 53.5 μg of luciferase/liver was obtained (FIG. 1B, PV+CL). The efficiency of gene transfer in the rat was compared to that in mice two ways. In terms of efficiency as defined by the mg of luciferase per mg of tissue weight, the levels of expression for portal injections were 3.6 μg of luciferase/g of tissue in rats as compared to 3.7 μg of luciferase/g of tissue in mice. Alternatively, in terms of efficiency as defined by the ng of luciferase per jig of pDNA delivered, the efficiencies of expression for portal injections were 71 ng of luciferase/μg DNA in rats as compared to 37 ng of luciferase/μg DNA in mice.[0149]

Less but still substantial luciferase expression was obtained when the injections of pCILuc were done into the efferent vessels of rats such as the IVC or bile duct (FIG. 1B). Injections of 750 μg of pCILuc in 15 ml into the hepatic vein (via the IVC) while occluding the portal vein yielded an average of 1.5 μg of luciferase/liver. Injections of 750 μg of pCILuc in 5 to 8 ml into the bile duct without any outflow obstruction yielded an average of 1.3 μg of luciferase/liver.[0150]

Parenchymal pressures of 12-50 mm Hg were measured in 23 rat livers during the injection of 750 μg pCILuc in 15 or 20 ml into the portal vein while occluding the IVC (FIG. 2). Spearman's rank correlation between pressure and luciferase expression was 0.76 (two-side p-value <0.001) indicating that pressure and luciferase were significantly positively associated. It appeared that pressures of over 40 mm Hg did not result in increased expression. The necessity of the non-linear component over and above a simple linear fit was verified by an approximate full versus reduced F-test (p-value =0.0 14), indicating that the observed plateau effect is real. Examination of both the residual quantile-quantile and the residual versus prediction plots further reveals that there are no serious violations of the regression model assumptions and therefore the regression models and p-values are valid (Fisher and van Belle, 1993).[0151]

Preliminary experiments explored the ability of naked pCILux (not pCELuc) to be expressed in dogs (Table I). In mice, in the case of intraportal injection, pCILux provided 3× lower expression than pCILuc (data not shown). In five dogs, various amounts of DNA were injected into either the bile duct with or without blocking outflow by occluding the IVC. In one animal, the DNA was injected into the IVC without any outflow blockage. All the dogs survived the procedure except animals that had the IVC occluded recovered more slowly post-operatively. The animals were sacrificed one day after the injections and dozens of tissue samples from each liver lobe were analyzed for luciferase. The luciferase expression was evenly distributed over all the lobes in each liver except in one lobe of one dog. Routine histological analysis in dog #5 (Table I) indicated that the tissue architecture was substantially disrupted suggesting that the injection volumes were too large.[0152]

Decreasing the volume of injection in[0153]dog #6 to 200 ml resulted in the best expression (Table I).

Mice and Rat β-Galactosidase Results: The β-galactosidase expression vector was used to determine the percent and type of cells that were transfected (FIG. 3). As previously noted for portal vein injections (Budker, et al., 1996), the vast majority of the blue-stained cells appeared to be hepatocytes on morphological grounds but a few appeared to be endothelial or other types of cells. A preponderance of hepatocytes were also stained blue after the bile duct or IVC injections in mice or rats (FIG. 3).[0154]

Under the various injection conditions, the percent of cells paralleled the levels of luciferase expression (FIGS. 1 and 3). In mice, the IVC injections gave the highest percentage of β-galactosidase-positive cells in which 7% of hepatocytes were positive (FIG. 3B). In rats, the portal vein injections gave the highest percentage of β-galactosidase-positive cells in which 7% of hepatocytes were positive (FIG. 3C). In some animals, liver cell damage was evident in less than 5% of the cells (for example in FIG. 3A). Of note in the rat livers injected into the portal vein, almost all of the positively-stained cells were periacinar with few positive cells around the central vein (FIG. 3C).[0155]

Repeat Bile Duct Injections: The bile ducts of mice were cannulated and 100 μg of pCMVhGH in 1 ml of 15% mannitol in normal saline were injected once a week (FIG. 4). Serum levels of hGH increased one day after the fast injection and then decreased to background levels by seven days after injection. One day after the second injection, hGH levels again increased and then were back to background levels by seven days after the second injection. Only minimal increases in hGH levels occurred after the third injection. Mice that had the highest levels after the first injection had the lowest levels after the second injection ([0156]mice 3 and 6) and vice versa (

mice

1, 2, and 4). In another set of animals (4 mice), the bile duct injections were repeated four times with pCMVhGH and then pCELuc was injected. The first three pCMVhGH injections led to similar increases in hGH serum levels as in FIG. 4. Although there were only minimal raises in hGH serum levels following the fourth injection, injection of pCILuc yielded an average of 29.2 ng/liver (±7.1, n=3). The liver in one of the four mice was grossly yellow and scarred as a result of the bile duct ligation and did not express any luciferase.

Discussion

This report extends the findings of the previous study showing pDNA expression following afferent intraportal delivery and demonstrates efficient plasmid expression following delivery via efferent vessels such as the hepatic vein or bile duct. Expression of luciferase or β-galactosidase was evenly distributed throughout the entire liver when either of the three vessels were injected. Combining these surgical approaches with improved plasmid vectors enabled uncommonly high levels of foreign gene expression in which over 15 μg of luciferase protein/liver was produced in mice and over 50 μg in rats (FIG. 1 and[0157]2). Equally high levels of β-galactosidase expression were obtained in that 7% of the hepatocytes had intense blue staining with only a one hour X-GAL incubation (FIG. 2). These levels of foreign gene expression are among the highest levels obtained with nonvital vectors and approach what can be achieved with vital vectors.

Using the portal vein administration route, occlusion of the outflow is critical for expression. Outflow occlusion increases the expression with the efferent administrative routes, but substantial amounts of expression were obtained even when the hepatic vein was not blocked. Most likely the natural direction of blood flow provides a sufficient impetus to retard the egress of injection fluid and raise the hydrostatic pressure. The use of these efferent vessels simplifies the administration for potential human applications since they are easier to access by non-invasive methods. If no occlusion is used then only one vessel has to be reached. These efferent routes should also be considered for the administration of viral and non-viral vectors as has been done with the delivery of adenoviral vectors into the bile duct [Vrancken Peeters, et al., 1996b; Vrancken Peeters, et al., 1996a; Yang et al., 1993).[0158]

The mechanism of pDNA uptake is not known but may involve native cellular uptake processes (Budker, et al., 1996). It is of interest that high levels of luciferase expression could occasionally be obtained when the DNA was injected into the bile duct in small volumes of isotonic solutions without occluding the IVC. Increased osmolar and hydrostatic pressure may not be critical for uptake of the pDNA by hepatocytes as they are not in muscle cells (Wolff, et al., 1990; Wolff, et al., 1991; Wolff, et al., 1992a). This would suggest that the mechanism of pDNA uptake may in fact involve endogenous cellular pathways. Increased hydrostatic and osmotic pressures may raise expression by enhancing these cellular internalization processes (Haussinger, 1996). For example, hepatocyte shrinkage stimulates replication of the duck hepatitis B virus (Offensperger, et al., 1994).[0159]

The raised pressures could also increase the delivery of the pDNA to the hepatocyte surface not only for the blood vessel administrations (through the sinusoidal fenestrae) but for the bile duct injections as well. The increased pressures could transiently attenuate bile secretion thereby decreasing the clearance of the pDNA (Stieger, et al., 1994). For example, the hyperosmolar mannitol should induce cell shrinkage that is known to inhibit taurocholate excretion into bile (Haussinger, et al., 1992).[0160]

Hepatocytes are functionally polarized cells in which the basal and apical membranes have different exocytic, endocytic and transcytotic functions (Hubbard, et al., 1994). The success with either blood vessel or bile duct routes could indicate that both the basal (sinusoidal) and apical (bile canalicular) membranes share a common pathway for the cellular entry of pDNA. However, the passage of DNA between the basal-lateral and bile canalicular spaces cannot be excluded even under the more gentle injection conditions. Plasmid DNA injected into the bile duct could be taken up from the basal-lateral surface of hepatocytes after passage pancellularly. If only the basal-lateral surface takes up the pDNA then the increased hydrostatic and osmotic pressures may enhance pDNA expression by disrupting the tight junctions between hepatocytes and thereby increase the flow of pDNA between the canalicular and basal-lateral spaces (Desmet and De Vos, 1982).[0161]

The bile duct was cannulated to determine whether repeat injections could be done. Substantial hGH levels were obtained alter the first two injections (FIG. 3). The hGH levels dropped considerably by one week after each injection presumably because of an immune response to the foreign protein. The inability to detect hGH after the third or fourth injections was most likely due to a more rapid response of the immune response sensitized to hGH. The ability to subsequently obtain luciferase expression argues against an immune response agamst the pDNA that prevents expression. Previous studies have failed to detect anti-DNA antibodies following the administration of naked pDNA (Nabel, et al., 1992; Jiao, et al., 1992). The ability to repetitively administer naked pDNA without inducing an immune response against the vector is a distinct advantage of naked pDNA over viral and some types of non-viral vectors.[0162]

Other studies in progress in our laboratory suggest that suppression of the immune system enables more persistent expression. In post-mitotic myofibers, plasmids can persist extrachromosomally and express for at least two years; presumably because the pDNA is not being lost as result of cell division (Wolff, et al., 1992b; Herweijer, et al., 1995). Quite possibly pDNA would be lost slowly in hepatocytes which have a half-life of up to a year in rodents and humans (Webber, et al., 1994; Leffert, et al., 1988). If so then liver-based genetic disorders such as hemophilia could be treated by injections every six months. The bile duct could be accessed repeatedly by upper gastrointestinal endoscopy. Similarly, the hepatic vein could be non-invasively accessed via peripheral or central veins. In addition, gene transfer could be delivered to newborns via the umbilical cord vessels to get them over a newborn metabolic crisis as occurs in the organic acidunas and the urea cycle defects.[0163]

Often gene transfer techniques that work in mice do not work in larger animals. Our results demonstrate that the technique works in rats that are approximately ten-fold bigger than mice. The dog results indicate that the liver of larger non-rodent mammal can express naked pDNA. Although substantial levels of luciferase activity were obtained, further optimization of the injection conclidons is required to increase the efficiency of expression so that they are comparable to those in rodents. The studies to determine the relationship between intraparenchymal pressure and luciferase expression in rats are a first step towards this goal. Minimal liver cell damage occurred in the rodents as evident by serum chemistries and histology but the injections were more disruptive to the hepatocytes in dogs. Presumably, the key factor is the efficient delivery of the pDNA to the hepatocyte surface with minimal cellular or tissue disruption (FIG. 2).[0164]

In the research laboratory, the described techniques will enable rodents to be used just as immortalized and primary liver cell cultures are now used for gene and cellular studies of liver function. The transfer of genes into cells in culture have been a critically important tool for deciphering the function of genes and for studying the effect of expressed proteins on cellular processes. Typically, the gene under study is placed within a plasmid vector and transiently transfected into the appropriate cell in culture. Isoforms and mutant forms of the gene under study can be quickly placed into plasmid expression vectors and studied. Our findings indicate that a similar plasmid-based approach could be used to study the effects of gene function in hepatocytes in situ. Given that the high levels of expression are transient in this system, it would be best if these effects occurred within a few days. The use of pDNA avoids the laborious steps necessary for the production of vital vectors or generation of transgenic mice and thereby enables many different genes and their related mutated forms to be quickly studied. It will permit are mechanism of gene expression and their effects on liver function to be expeditiously probed within the context of a complete mammalian organism.[0165]

Example 12

The intravascular delivery of naked pDNA to muscle cells also has attraction. Muscle has a high density of capillaries (Browning, 1996) and the capillaries are in close contact with the myofibers (Lee 1995). However, the endothelium in muscle capillaries is of the continuous, non-fenestrated type and has low solute permeability especially to large macromolecules(Taylor, A. E., and D. N. Granger. Exchange of macromolecules across the microcirculation. In: Handbook of Phusiology. The Cardiovascular System. Microcirculation..Bethesda, Md.: Am. Physiol.Soc., 1984, sect. 2, vol. IV, chapt. 11, p. 467) The mechanism of macromolecule transendothelial transport is poorly understood. Cell biologists have proposed that this transport occurs by transcytosis involving plasmalemmal vesicles or convective transport through transient transendothelial channels formed by the fusion of vesicles. (Michel C. C. Cardiovascular Res. 1996). Physiologists have modeled the muscle endothelium as having a large number of small pore with radii of 4 nm and a very low number of large pores with radii of 20-30 nm.(Rippe B. Physiological Rev, 1994). Although the radius of gyration of 6 kb pDNA is ˜100 nm (Fishman, D. M., Biopolymers, 38, 535-552), supercoiled DNA in plectonomic form has superhelix dimensions of approximately 10 nm (Rybenkov V. V. J. Mol. Biol. 267, 299-311 1997). This implies that pDNA has a possibility to traverse microvascular walls by stringing through their large pores. We hypothesized that the rate of pDNA extravasation could be increased by enhancing fluid convection through these large pores by raising the transmural pressure difference in selective regions. This report demonstrates that intravascular pDNA injections under high pressure can in fact lead to high levels of foreign gene expression in muscles throughout the selected hindlimb of an adult rat.[0166]

Hydrostatic Pressure

475 ug of pCILux in normal saline solution (NSS) was injected into the femoral arteries of adult Sprague-Dawley rats while blood inflow and outflow were blocked. Injection of pCILux, a luciferase expression vector utilizing the CMV promoter (LIVER II promoter accepted for HGT), was done after both the femoral artery and vein was occluded for 10 min. Two days after pDNA injections, the luciferase activities were measured in all the muscles of the hindlimb (FIG. 5). The highest levels of luciferase expression were achieved when the pCILux was injected in 9.5 ml of normal saline within 10 sec. Injection volumes less than 9.5 ml resulted in substantially less expression (FIG. 5([0167]a)). Injection times more than 10 sec also resulted in much less expression (FIG. 5(b)). This critical dependence on volume and speed of injection indicates that either increased hydrostatic pressure or rapid flow is required for efficient expression. Artery injections performed without occluding the femoral vein resulted in approximately 200-fold less expression (1.8±1.2 ng of luciferase per all hindlimb muscles).

Other Factors

Further studies were performed to determine the effect of ischemia on the expression of the intravascularly injected pCILux. The time of ischemia was adjusted by varying the time that both the femoral artery and vein were occluded prior to pDNA injection (FIG. 5([0168]c)). Although the highest level of expression was obtained with 10 min of ischemia, the expression levels were only a few-fold lower at shorter or longer ischemia times. This suggests that ischemia is not a critical factor for enabling egress of the pDNA out of the intravascular space. However, the blood flow for the zero ischemia time point is disrupted for 30 sec and this may be sufficient to affect vascular permeability. Ischemia could increase expression either by capillary recruitment and vasodilatation) or augmenting permeability (Mathieu-Costello O.Int J Microcirc 1995,15, 231-237). In addition, ischemia could possibly increase pDNA expression by affecting transcription or translation. Ischemia can be tolerated by muscle for two to three hours (Gidlof A.,Int. J. Microcirc. Clin. Exp. 1987, 7,67-86). Histologic analysis of the muscle did not reveal any hemorrhage or myofiber damage. Other factors were explored to increase the level of expression. Hypotonic (FIG. 6, condition 2) or hypertonic (FIG. 6, condition 3) injection solutions resulted in less expression. Although, the effect of the hypertonic injection solution ( normal saline solution with 15% mannitol) may have been mitigated by the slower rate of injection (over 30 sec instead of 10 sec) that results from its increased viscosity. The pre-injection with collagenase resulted in a 3.5-fold increase to 1,332 ng/total muscles (FIG. 6, condition 4). The collagenase was used to disrupt the basement membrane of the capillaries and thereby increase pDNA egress. It could have also increased expression by disrupting the extracellular matrix within the muscle. Further studies are in progress to determine the optimal conditions for collagenase administration without causing hemorrhage.

Distribution of Foreign Gene Expression

The luciferase expression was distributed among all muscle groups in the leg (Table 1). The lower levels in the lower anterior leg and foot are probably due to their high content of tendons and small amount of muscles. The levels after intravascular injection was up to 40 times higher than the levels after intramuscular injection.[0169]

The type and percentage of the transfected cells in the muscle were determined using the β-galactosidase reporter system. Using the best injection condition ([0170]condition 4 in FIG. 6), approximately 10-50% of the myofibers expressed β-galactosidase.

These expression results provide indirect evidence that pDNA extravasation occurred. More direct evidence was obtained using fluorescently-labeled DNA injected into the femoral artery ([0171]condition 1 in FIG. 6). The labeled DNA was distributed extravascularly in all the limb muscles and surrounded most of the myofibers (data not shown).

Conclusion

These results demonstrate that the intraarterial delivery of pDNA to muscle can be greatly enhanced when injected rapidly, in a large volume and when all blood vessels leading into and out of the hindlimb are occluded. These conditions presumably increase the intravascular hydrostatic pressure and thereby increase the convective outflow that transports the pDNA into contact with myofibers. The high intravascular pressure may increase the number, size and permeability of the microvascular pores (Rippe, B. Acta Physiol. Scand. 125, 453-459, 1985) (Wolf, M. B. Am. J. Physiol. 257, H2025-H2032,1989). Preliminary studies using collagenase suggest that enzymatic disruption of the vessels basement membrane or muscle extracellular matrix may also increase the delivery of pDNA to the myofibers. Ischemia also increased expression moderately. Furthermore, it is expected that other enzymes such as hylaronidase will perform similarly to collagenase.[0172]

This study shows that the intravascular approach increases expression in muscle up to 40-fold as compared to intramuscular injection. This is the first demonstration that naked plasmid DNA can be efficiently expressed in muscles of adult animals larger than mice. It is also the first report of an intravascular non-viral gene transfer approach for muscle. Further studies in larger animals will determine the clinical relevance of this study.[0173]

Description of Injections

A 4 cm long abdominal midline excision was performed in 150-200 g, adult Sprague-Dawley rats anesthetized with 80 mg/mg ketamine and 40 mg/kg xylazine. Microvessel clips (Edward Weck, Inc., Research Triangle Park, N.C.) were placed on external iliac, caudal epigastric, internal iliac and deferent duct arteries and veins to block both outflow and inflow of the blood to the leg. A 27 g butterfly needle was inserted into the external iliac artery and the DNA solution was injected by hand.[0174]

Table 1. Distribution of luciferase expression among the different hindlimb muscle groups two days following intraarterial injection with 475 μg of pCILux in 9.5 ml of NSS over 10. Means and their standard deviations are shown for 6 animals.[0175]



		Total Luciferase	Luciferase Conc.
Injection type	Muscle group	(ng)	(ng/g tissue)

Intravascular	Upper Leg Anterior	149.1 ± 51.2	109.2 ± 37.5
	Upper Leg Posterior	74.6 ± 36.0	43.4 ± 20.9
	Upper Leg Medial	88.7 ± 63.7	61.6 ± 44.5
	Lower Leg Posterior	114.5 ± 89.7	66.0 ± 51.7
	Lower Leg Anterior	3.0 ± 1.0	5.4 ± 1.7
	Foot	0.5 ± 0.2	4.1 ± 1.4
Intramuscular	Upper Leg Anterior	3.7 ± 0.9	2.8 ± 0.8

Example 13

Adenoviral vectors can be delivered to muscle parenchymal cells by an intravascular route.[0176]

Methods: the same methods that were used to deliver naked plasmid DNA to muscle in rats were also used. The adenoviral vector CMVLacZ that expresses the E. coli beta-galactosidase from the immediate early promoter of the human cytomegalovirus (commonly called the “CMV promoter”) was prepared as previously described (Yang, T., K. U. Jooss, Q. Su, H. C. J. Ertl and J. M. Wilson:[0177]Immune responses to viral antigens versus transgene product in the elimination of recombinant adenovirus-infected hepatocytes in vivo,Gene Therapy, 3(2): 137-144, 1996) The rat iliac artery was preinjected with 0.5 mg of papaverine and 40 ng of collagenase in 3 ml of saline while blocking the iliac artery and vein. 5×10⁸particles of the adenoviral vector CMVLacZ in 10 ml of saline was injected in about 10 sec and 2 minutes later the clamps were opened. Two days after injection the leg muscle were assayed for luciferase as above.

Results

Table: Distribution of luciferase activity following the intraarterial injection of adenovirus CMVLacZ.[0178]



	Muscle Group	Luciferase (ng)

	Upper Leg Anterior	59.04
	Upper Leg Posterior	18.33
	Upper Leg Medial	4.44
	Lower Leg Posterior	11.04
	Lower Leg Anterior	5.33
	Foot	0.22
	Total	98.40

The foregoing is considered as illustrative only of the principles of the invention. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation shown and described. Therefore, accordingly, all suitable modifications and equivalents fall within the scope of the invention.[0179]