¹Biotechnology Unit, Dibit, Department of Biological and Technological Research, San Raffaele Scientific Institute, Milan, Italy

²Department of Chemistry and Biochemistry, Medical School, University of Milan, Italy

³Tiget, Telethon Institute of Gene Therapy, Milan, Italy

Correspondence to: L Monaco, Dibit, San Raffaele Scientific Institute, Via Olgettina, 58, 20132 Milan, Italy

Gene transfer to the kidney can be achieved with various DNA vectors, resulting in transgene expression in glomerular or tubular districts. Controlling transgene destination is desirable for targeting defined renal cells for specific therapeutic purposes. We previously showed that injection of polyplexes into the rat renal artery resulted in transfection of proximal tubular cells. To investigate whether this process involves glomerular filtration of the DNA-containing particles, fluorescent polyethylenimine polyplexes were prepared, containing fluoresceinated poly-L-lysine. This allowed visualization of the route of the particles into the kidney. Our polyplexes were filtered through the glomerulus, since fluorescent proximal tubuli were observed. Conversely, fluorescent lipopolyplexes containing the cationic lipid DOTAP were never observed in tubular cells. Size measurements by laser light scattering showed that the mean diameter of polyplexes (93 nm) was smaller than that of lipopolyplexes (160 nm). The size of the transfecting particles is therefore a key parameter in this process, as expected by the constraints imposed by the glomerular filtration barrier. This information is relevant, in view of modulating the physico-chemical properties of DNA complexes for optimal transgene expression in tubular cells. Gene Therapy (2000) 7, 279-285.

kidney; non-viral gene delivery; fluorescent complexes: glomerular filtration barrier; laser light scattering; -galactosidase

Genetic renal diseases could in principle be corrected by somatic gene therapy. Different districts of the kidney are affected by different pathologies; for example, autosomal dominant polycystic kidney disease causes the formation of liquid-filled cysts along the whole tubule,¹ while Alport syndrome affects the glomerulus.² Therapeutic gene transfer must therefore be tailored to ensure expression of the transgene in the appropriate cell type. A number of viral, as well as non-viral, approaches have been used to transfer genes to the kidney, leading to expression of the transgene in the tubule,³ in the glomerulus,^4,5 or in renal vessels.⁶ In these cases, access to the kidney was achieved via direct injection into the left renal artery, but intrapelvic injections have also been employed, leading to transgene expression in the papilla and medulla,³ or in the outer medulla.⁷ In our previous work, we showed that DNA complexed with branched 25 kDa PEI at 10 equivalents (eq) of amino to phosphate (N/P) groups is able to transfect proximal tubular cells following injection into the rat renal artery. Conversely, transfection efficiency of DNA complexed to 5 N/P eq of PEI, or to cationic lipids such as DOTAP, was significantly lower.⁸ Given the anatomical complexity of the kidney in the present work we investigated the route of these complexes in this organ, in an attempt to correlate the physico-chemical properties of these particles to tissue localization and ultimately to transfection efficiency.

DNA complexes entering the kidney from the vasculature could have access to tubular cells by two alternative routes. The first one involves the passage into the urinary space through the multiple layers constituting the glomerular filtration barrier. Alternatively, DNA complexes could exit the glomerular vascular system, travel to peritubular capillaries and hence pass into tubular cells (Figure 1). Definition of the properties of the complexes affecting the choice of either route would offer a rational basis for the design of suitable non-viral vectors for the kidney. Fluorescently tagged DNA complexes can be used to follow the distribution of such particles within organs, following systemic administration into laboratory animals,⁹ but no detailed information on the kidney was available so far. We chose fluorescein isothiocyanate-conjugated poly-L-lysine (FITC-pLys) as a convenient, commercially available fluorophore for either polyplexes or lipopolyplexes. Fluorescent complexes were prepared by partial pre-condensation of DNA with 0.4 N/P eq of FITC-pLys, followed by complete condensation with either PEI, to yield polyplexes, or with DOTAP, to yield lipopolyplexes. Inclusion of pLys into lipoplexes to yield lipopolyplexes has been described to enhance transfection efficiency in cultured cells.¹⁰ FITC-pLys-containing lipopolyplexes have been used to detect DNA particles in vitro and in vivo.¹¹ The preparation of mixed polyplexes, including both FITC-pLys and PEI, has not been reported so far. In the present study, we made no effort to increase transfection efficiency by optimizing the amount of FITC-pLys; rather, the FITC-pLys to DNA ratio was kept low enough to give only partial condensation of DNA,¹¹ as confirmed by our laser light scattering measurements (see below), but sufficient to allow observation of the complexes at the fluorescence microscope. Complete condensation of DNA was ensured by the addition of either PEI or DOTAP, at concentrations known to yield fully condensed DNA complexes in the absence of polylysine (pLys). We reasoned that the properties of the complexes would be only minimally affected by such a low amount of pLys. Quality control experiments were performed, to assess whether the additional component in the fluorescent complexes would significantly affect the behavior of our complexes in terms of transfection efficiency. Transfection of CHO cells with a luciferase reporter plasmid complexed with either 10 or 5 PEI eq and FITC-pLys yielded transfection efficiencies comparable to regular DNA/PEI complexes (data not shown); inclusion of FITC-pLys into DOTAP complexes yielded an unexpected reduction of approximately 10-fold in transfection efficiency that was not observed in other cell lines (human embryonic kidney, HEK 293, and human type II pneumocytes, A549, data not shown). No attempt was made to clarify this phenomenon further as it was outside the scope of this work. On the other hand, since cultured cells are not bona fide models for in vivo experiments, transfection efficiency of the luciferase DNA complexes prepared with or without FITC-pLys and either PEI or DOTAP was also assessed in vivo by injection into the rat left renal artery. No significant differences were observed following inclusion of FITC-pLys between groups of animals treated with either DNA/PEI complexes or DNA/DOTAP complexes (Figure 2).

To investigate the fate of the injected complexes in the kidney, FITC-labeled polyplexes at 10 N/P eq of PEI were injected into the rat left renal artery. Animals were nephrectomized immediately after a period of ischemia lasting 10 min to allow polyplexes contacting the kidney, or at different time points that extended up to 30 min following restoration of the circulation. Groups of four animals were treated for each condition. The distribution of complexes described below was consistent within each group. In sections from kidneys at t = 0, fluorescence was observed in cortical areas, where the glomerular tuft was intensely stained; moreover, the lumen of some proximal tubuli was also fluorescent. While fluorescence intensity was comparable in all of the stained glomeruli, completely dark glomeruli could also be identified (Figure 3a). This observation points to an uneven distribution of the exogenous genetic material within the organ, a possible crucial reason for the reduced number of nephrons ultimately displaying expression of the transgene (see below). Indeed, recirculation of recombinant adenovirus within pig kidneys for at least 30 min was required for expression of the reporter gene for -galactosidase in an elevated proportion of glomeruli.⁵ Our observation confirms that technical optimization of the delivery of non-viral, as well as of viral vectors through the renal artery is required for enhanced transfection efficiency in the kidney. In some sections of kidneys injected with the FITC-pLys/PEI polyplexes at 10 N/P eq, the endothelium of arcuate and medium arteries was also occasionally labeled (data not shown). Fluorescence in a glomerulus and in the corresponding tubule was visible in some sections (Figure 3b), clearly showing the continuous passage of fluorescent particles through the glomerular filtration barrier. Two minutes after ischemia, the glomerular staining was slightly less intense, while a higher number of stained proximal tubuli was detected, as compared with sections prepared at t = 0. Indeed, the tubular staining often involved the glomerular urinary pole, as well as more distant proximal tubular portions (Figure 3c). Confocal fluorescence analysis confirmed the above observation; in particular, Figure 3d shows staining of the tubular brush border, supporting the access of complexes to tubular cells from the luminal side. Downstream of the filtration barrier, the fluorescent signal was essentially present in proximal tubuli, to indicate a rapid uptake of the complexes by cells along the initial tract of the tubule. At longer times (5 and 10 min after ischemia), a general decrease in the intensity of fluorescence was observed, with no significant alteration in the distribution of the signal. Complete disappearance of the signal was observed at t = 30 min (data not shown). Overall, the fluorescent signal was observed mainly in the medial region of the kidney immediately after the ischemic period, while it involved also the kidney poles 2 min later.

Next, we wanted to confirm that the FITC-signal belonged to the DNA-containing complexes. Therefore, plasmid DNA was labeled by covalent binding to ethidium monoazide (EMA), and doubly labeled polyplexes (including EMA-DNA, 0.4 N/P eq of FITC-pLys and 10 N/P eq of PEI) were prepared. In preliminary experiments on cultured CHO cells, co-localization of the DNA and pLys labels was observed 1 h after transfection (data not shown). These doubly labeled polyplexes were also injected into the kidney, and co-localization of the DNA signal with FITC-pLys was observed at t = 2 min in glomeruli as well as in tubuli (Figure 3e-h). A faint, diffuse staining was often visible in kidney sections at both the FITC and EMA wavelengths; this was due to autofluorescence of the kidney, as shown in sections from control animals injected with non-fluorescent complexes (polyplexes prepared with 10 N/P eq of PEI, 0.4 N/P eq of pLys and non-labeled plasmid DNA), as shown in Figure 3j-k.

Having established that fluorescent PEI polyplexes are able to reach the tubular lumen through a filtration mechanism, we wanted to investigate the fate of positively charged lipopolyplexes prepared with a different synthetic vector, namely the cationic lipid DOTAP. We observed (Figure 1b) that luciferase transfection efficiency of DOTAP complexes in the kidney was lower than that of PEI complexes, in keeping with our previous results.⁸ Fluorescent DOTAP lipopolyplexes were detected in glomeruli and in small and medium arteries when injected by the above procedure, but were completely absent in tubuli at both the cortical (Figure 3k) and medullary level.

Our observations about the selective behavior of DNA complexes within the glomerulus needed to be correlated with the dimensions of the complexes themselves, in consideration of the size limitations dictated by the glomerular filtration barrier.¹² Results from dynamic laser light scattering measurements are summarized in Figure 4. The means of the mean particle size ± s.e. are reported (n = 3-5, see below). 0.4 N/P FITC-pLys/10 N/P PEI polyplexes showed an average particle size of 93 ± 13 nm (n = 4), while DNA/FITC-pLys/DOTAP lipopolyplexes were significantly larger, with a mean size of 160 ± 1 nm (n = 4). The different route of these complexes in the kidney can therefore be accounted for by the observed difference in size. The size of the fenestrae in the glomerular endothelium is on average 70 nm, and the generally accepted glomerular filtration cutoff for cationized proteins (eg ferritin) is 14 nm.¹² The observed filtration of the PEI polyplexes are in contrast with this information, and might be explained by the non-globular shape or by the deformability of the complexes, an issue which has not been investigated so far. In addition, it is possible that only the smaller particles within the polyplex population are able to pass the glomerular filtration barrier. Interestingly, transgene expression in proximal tubular cells following arterial injection of recombinant adenovirus (70-100 nm in size) has been described.³

To understand the influence of the pre-condensing agent FITC-pLys on the size of the complexes, the dimensions of fluorescent complexes were compared with those of the corresponding complexes prepared without the fluorophore. PEI polyplexes at 10 N/P eq were slightly smaller (89 ± 9 nm, n = 5), while DOTAP lipoplexes were slightly larger (187 ± 10 nm, n = 4) than the corresponding FITC-pLys-containing complexes. In both cases, these differences were not statistically significant, indicating that FITC-pLys did not grossly affect the overall particle dimension in these cases. The size of the doubly labeled polyplexes (105 ± 3 nm, n = 3) was also statistically comparable to that of the other polyplexes at 10 N/P PEI eq (Figure 4). FITC-pLys as such was not able to condense DNA at dimensions below 10 m (not shown on Figure 4).

Both the PEI polyplexes and the DOTAP lipopolyplexes described above bear a nominal positive charge, with a calculated +/- charge ratio of 2.1 and 3.0, respectively. In the case of PEI, one out of six nitrogen atoms was considered as positively charged at physiological pH.¹³ Since passage through the glomerular filtration barrier depends on both the size and the charge of the particles,¹² we wished to investigate how modulation of the charge parameter in DNA complexes would affect access of the exogenous genetic material to the lumen of the nephron. 5 N/P PEI/0.4 N/P FITC-pLys complexes were prepared, corresponding to a slightly positive nominal charge ratio of 1.2. The resulting fluorescence pattern at t = 2 min was essentially glomerular, as shown in Figure 3l. It was important to elucidate whether the effect of the complex charge on the filtration process was independent from the size parameter. The size of such polyplexes was larger than 10 m (n = 4), while the size of the corresponding polyplexes prepared in the absence of FITC-pLys was statistically comparable (77 ± 15 nm, n = 4) to that of the PEI-containing complexes shown in Figure 4. We postulate that the effective +/- charge ratio of the 0.4 eq FITC-pLys/5 eq PEI complexes is equal or close to 1.0. This would abolish the repulsive interactions among particles and lead to the formation of aggregates. Indeed, isoelectric complexes (where +/- equals 1.0) have been described to yield large aggregates.^14,15 Unfortunately, we were unable to identify formulations yielding small-sized, low-charged fluorescent polyplexes, which would allow us to analyze the involvement of the charge in the process of glomerular filtration of the complexes. In fact, heterogeneous populations of particles were obtained for complexes prepared with PEI in the range of 3 to 7 eq and 0.2 to 0.8 eq of FITC-pLys (data not shown). Clearly, the interactions among the three components of these mixed polyplexes are very complex within this range of relative concentrations: further work would be needed to elucidate such interactions at the structural level.

To confirm that the physical distribution of the DNA complexes in the injected kidney co-localized with the site of expression of the transgene, the gene for bacterial -galactosidase was used as a reporter. The resulting polypeptide was directed to the nucleus by using a modified lacZ cDNA coding for a chimeric -galactosidase equipped with a nuclear localization signal. This allowed a clear-cut identification of cells expressing the transgene vs cells displaying endogenous -galactosidase activity in the cytoplasm. Proximal tubular cells positive for nuclear -galactosidase were observed in sections from left kidneys transfected with DNA/FITC-pLys/PEI complexes at 10 PEI N/P eq (Figure 5a-c), in keeping with our previous observation.⁸ No stained nuclei were found in either kidney of rats transfected with an irrelevant plasmid (Figure 5d and e), nor in the right kidneys of treated animals (Figure 5f). Only a small proportion of proximal tubular cells displayed -galactosidase activity: this could be explained by our observation that only a limited number of nephrons did get in physical contact with polyplexes. Moreover, the use of -galactosidase as a reporter gene could lead to an underestimation of the number of cells which were effectively transfected, due to the limited sensitivity of the assay.¹⁶ In addition, we had to modify the incubation parameters for X-gal, in order to reduce the non-specific signal due to endogenous -galactosidase activity in kidney cells. This could further reduce the sensitivity of the detection method. Although the fluorescent signal of polyplexes co-localized with glomerular cells, and was present also on the endothelium of arterioles, no -galactosidase activity was ever detected in such cells. This confirms more general observations leading to the concept that internalization of DNA complexes is required but not sufficient to elicit expression of the transgene.⁹ In kidneys transfected with DOTAP/FITC-pLys/DNA complexes, no nuclear -galactosidase activity was detected in tubular nor in glomerular cells. Kidneys transfected with such complexes displayed measurable luciferase activity, which was on average five-fold lower than the activity elicited by the DNA/FITC-pLys/PEI complexes at 10 N/P eq of PEI (Figure 2). Nonetheless, we expected to identify at least a few -galactosidase-positive cells following transfection with the lacZ lipopolyplexes, but were unable to do so at either the glomerular or tubular level. One possible reason for this could be the limited sensitivity of the detection assay, as discussed above. Of note, infarcts were frequently observed in DOTAP-transfected kidneys. Only a few cells in the interstitium surrounding arteries of the renal ileum were positive for -galactosidase, in areas close to the infarctions (data not shown). All of these experiments were repeated at least twice for confirmation.

In conclusion, we have established that polyplexes able to transfect proximal tubular cells have access to these cells through glomerular filtration. Consequently, a careful design and characterization of the physico-chemical properties of DNA particles is required to have access to defined districts of the kidney.

Acknowledgements

This work was supported by the Italian Telethon grant A.112. The authors wish to thank Dr Umberto Fascio (Milano) for hospitality and support for the confocal microscopy observations, Professor Mario Corti (Milano) for encouragement in the laser light scattering work and Professor Daniele Cusi (Milano) for helpful discussions on the intricacies of glomerular filtration.

1 Gabow PA. Autosomal dominant polycystic kidney disease. New Engl J Med 1993; 329: 332-342, MEDLINE

2 Kashtan CE. Alport syndrome and thin glomerular basement membrane disease. J Am Soc Nephrol 1998; 9: 1736-1750, MEDLINE

3 Moullier P et al. Adenoviral-mediated gene transfer to renal tubular cells in vivo. Kidney Int 1994; 45: 1220-1225, MEDLINE

4 Isaka Y et al. Glomerulosclerosis induced by in vivo transfection of transforming growth factor-beta or platelet-derived growth factor gene into the rat kidney. J Clin Invest 1993; 92: 2597-2601, MEDLINE

5 Heikkila P et al. Adenovirus-mediated gene transfer into kidney glomeruli using an ex vivo and in vivo kidney perfusion system - first steps towards gene therapy of Alport syndrome. GeneTherapy 1996; 3: 21-27,

6 Zhu G et al. In vivo adenovirus-mediated gene transfer into normal and cystic rat kidneys. Gene Therapy 1996; 3: 298-304, MEDLINE

7 Lai LW, Moeckel GW, Lien YH. Kidney-targeted liposome-mediated gene transfer in mice. Gene Therapy 1997; 4: 426-431, MEDLINE

8 Boletta A et al. Nonviral gene delivery to the rat kidney with polyethylenimine. Hum Gene Ther 1997; 8: 1243-1251, MEDLINE

9 McLean JW et al. Organ-specific endothelial cell uptake of cationic liposome-DNA complexes in mice. Am J Physiol 1997; 273: H387-H404, MEDLINE

10 Gao X, Huang L. Potentiation of cationic liposome-mediated gene delivery by polycations. Biochemistry 1996; 35: 1027-1036, Article MEDLINE

11 Vitiello L et al. Condensation of plasmid DNA with polylysine improves liposome-mediated gene transfer into established and primary muscle cells. Gene Therapy 1996; 3: 396-404, MEDLINE

12 Kanwar YS et al. Current status of the structural and functional basis of glomerular filtration and proteinuria. Semin Nephrol 1991; 11: 390-413, MEDLINE

13 Suh J, Paik HJ, Hwang BK. Ionization of poly(ethylenimine) and poly(allylamine) at various pHs. Bioorg Chem 1994; 22: 318-327,

14 Lee RJ, Huang L. Folate-targeted, anionic liposome-entrapped polylysine-condensed DNA for tumor cell-specific gene transfer. J Biol Chem 1996; 271: 8481-8487, MEDLINE

15 Kwoh DY et al. Stabilization of poly-L-lysine/DNA polyplexes for in vivo gene delivery to the liver. Biochim Biophys Acta 1991; 1444: 171-190,

16 Couffinhal T et al. Histochemical staining following LacZ gene transfer underestimates transfection efficiency. Hum Gene Ther 1997; 8: 929-934, MEDLINE

17 Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press: New York, 1989,

18 Monaco L et al. An in vitro amplification approach for the expression of recombinant proteins in mammalian cells. Biotechnol Appl Biochem 1994; 20: 157-171, MEDLINE

19 Mohr L et al. Targeted gene transfer to hepatocellular carcinoma cells in vitro using a novel monoclonal antibody-based gene delivery system. Hepatology 1999; 29: 82-89, MEDLINE

20 Corti M. Physics of Amphiphiles: Micelles, Vesicles and Microembulsions. North-Holland: Amsterdam, 1985,

21 Monaco L. Optimising gene expression in animal cells. In: Jenkins N (ed.). Mammalian Cell Biotechnology. Humana Press: Totowa, NJ, 1999, 39-48.

Figure 1 Schematic representation of the nephron, illustrating the possible routes followed by DNA complexes injected into the renal artery to reach tubular cells, where expression of the transgene has been observed.⁸ Complexes enter the glomerulus via the afferent arteriole, reach the glomerular tuft, and (A) hence could move into the urinary space through the glomerular filtration barrier. Alternatively (B), DNA complexes could be retained within the vasculature, exit the glomerulus via the efferent arteriole and eventually reach tubular cells from peritubular capillaries.

Figure 2 Comparison of fluorescent versus non-fluorescent DNA complexes in transfection experiments in vivo. Rat left kidneys were injected through the renal artery with luciferase DNA complexes prepared either with or without 0.4 N/P eq of FITC-pLys and with 10 N/P eq of PEI, or with DOTAP at a 6 w/w excess. Luciferase activity was measured in supernatants of kidney homogenates 48 h after transfection and is expressed as RLU/kidney. P, PEI; D, DOTAP; FL, FITC-pLys. Means ± s.d., as well as individual values, are reported for each group of animals. Bars with the same filling are not significantly different, bars with different fillings are significantly different among one another (P < 0.05), as established by one-way analysis of variance (ANOVA); multiple comparisons were performed using Fisher PLSD. The complexes used in these experiments were prepared with plasmid pCLuc, carrying the P. pyralis luciferase coding region under the control of the cytomegalovirus (CMV) immediate-early enhancer/promoter region (a kind gift of Dr S Chiocca, Institute of Molecular Pathology, Vienna). All plasmid DNA preparations were performed by double CsCl gradient purification.¹⁷ For complex preparation, a 100 mM aqueous stock solution of 25 kDa PEI (Aldrich, Milwaukee, WI, USA) was prepared as described.⁸ DOTAP (N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulphate, Boehringer Mannheim, Roche Molecular Biochemicals, Mannheim, Germany) was obtained as a 1 mg/ml aqueous solution. 16.1 kDa FITC-conjugated poly-L-lysine hydrobromide (Sigma, St Louis, MO, USA) was prepared as 1 mM stock solution in water. PEI-containing DNA complexes were formed in 150 mM NaCl, DOTAP-containing complexes in 20 mM Hepes buffer, pH 7.4, as described.⁸ FITC-pLys-containing complexes were prepared by pre-incubation of 0.4 N/P eq of FITC-pLys with DNA for 15 min at room temperature, followed by addition of either PEI or DOTAP. The in vivo transfection of rat kidneys was performed by injection into the left renal artery, as described.⁸ Eight-week-old male Sprague-Dawley rats were used. Complexes were kept in contact with the kidney for 10 min, before restoring circulation. Forty-eight hours later, animals were killed and kidneys were collected as described.⁸ Transfected kidneys were cut into pieces, homogenized on ice in 4.5 ml lysis buffer (25 mM Tris HCl, 2 mM dithiothreitol, 2 mM ethyilenediaminetetraacetic acid, 10% glycerol, 1% Triton X-100) and centrifuged for 10 min at 4°C at 14 000 g in a microcentrifuge. Luciferase assay was performed on supernatants using a luciferase assay kit by Promega (Madison, WI, USA). Luciferase activity was measured in a single-well luminometer (Lumat LB 9501, EG&G Bertholdt, Wellesley, MA, USA) for 30 s, and expressed as RLU per whole organ.

Figure 3 Distribution of fluorescent DNA complexes injected into rat kidneys. Complexes were prepared with 100 g per animal of plasmid pSfiSV19,¹⁸ 0.4 N/P eq of FITC-pLys and either PEI or DOTAP, and injected into the rat left renal artery as described in the legend to Figure 2. For doubly labeled complexes, DNA was labeled with EMA as described.¹⁹ Kidneys were removed immediately after the 10 min of ischemia (t = 0), or 2 min later (t = 2 min), fixed and sectioned. (a) Fluorescent polyplexes containing 10 N/P eq of PEI, t = 0. Intense staining is visible in the glomerular tuft (small arrows) and in the lumen of a few proximal tubuli (large arrows); dark areas correspond to non-fluorescent glomeruli (arrowhead). (b) Higher magnification of a section from a kidney treated as in panel a, showing a fluorescent glomerulus and the corresponding fluorescent tubule. (c) Fluorescent polyplexes containing 10 N/P eq of PEI, t = 2 min. Fluorescent glomeruli and still more fluorescent tubuli, as compared to panel a, are observed. (d) A section from the same kidney as in panel c was observed at the confocal microscope and shows staining of the tubular brush border. (e-h) Doubly labeled polyplexes at 10 N/P eq of PEI, t = 2 min. Sections were observed at the FITC wavelength (e and g) or at the EMA wavelength (f and h). Glomerular (small arrows) and tubular (large arrows) colocalization of the FITC and EMA labels is indicated. (i and j) Non-fluorescent polyplexes containing 10 N/P eq of PEI, 0.4 N/P eq of pLys (18 kDa poly-L-lysine, Sigma), t = 2 min. A background staining is visible. (k) Fluorescent lipopolyplexes prepared with DOTAP, t = 2 min. Fluorescent glomeruli as well as arterioles (arrowheads) are visible. (l) Fluorescent polyplexes containing 5 N/P eq of PEI, t = 2 min. The staining is almost exclusively glomerular (arrow); the fluorescent epithelium of a Bowman capsule (large arrow) and fluorescent arterioles (arrowhead) are also visible. Bar corresponds to 90 m for panels a, c, k and l; 37.5 m for panels e, f, i and j; 13.5 m for panels b, g and h; 9.4 m for panel d. For cryosection preparation, kidneys were removed and cut into 1-2-mm thick slices. Samples were fixed in 4% paraformaldehyde in PBS at 4°C overnight, rinsed three times in PBS, cryoprotected in 10% sucrose in PBS at 4°C overnight and frozen in liquid nitrogen. Samples were kept in the dark throughout all washing and incubation steps. Samples were then included in tissue freezing medium (TFM, Electron Microscopy Sciences, Fort Washington, PA, USA) and 5- or 20-m thick cryosections were prepared for observation at the fluorescence microscope (Axioplan; Zeiss, Oberkochen, Germany) or at the confocal microscope (TCSNT confocal laser scanning microscope; Leica Microsystems, Wetzlar, Germany), equipped with a laser argon/krypton 75 mW multiline. Confocal microscopy was performed on focal series of horizontal section planes, either monitored for FITC only, using the 488 laser line and an LP 515 filter, or simultaneously monitored for FITC and EMA using the 488 and 568 nm laser lines, a band pass 530/30 filter for FITC and a long pass 590 filter for EMA.

Figure 4 Size and kidney localization of DNA complexes. The size of the fluorescent complexes used for the in vivo experiments shown in were compared with the size of their non-fluorescent counterparts; P, 10 eq of PEI; D, DOTAP in 6 w/w excess over DNA; FL, 0.4 eq of FITC-pLys; EMA, EMA-labeled DNA. Bars represent the means ± s.e. of the mean particle size of repeated measurements (P, n = 5; P/FL, P5, D and D/FL, n = 4; P/FL/EMA, n = 3). Bars with the same filling are not significantly different, bars with different fillings are significantly different among one another (P < 0.05); statistical analysis was performed as described in the legend to Figure 2. DNA complexes were prepared as described in the same legend and diluted to 1.7 g DNA/ml in 150 mM NaCl for PEI complexes or in 20 mM Hepes buffer, pH 7.4 for DOTAP complexes. Dynamic laser light scattering measurements were performed at 19°C in an apparatus including a He-Ne laser, a temperature controlled cell and a digital correlator (Brookhaven Instruments model 2030, Holtsville, NY, USA), as described.²⁰ To exclude the presence of interparticle interactions, which could affect the results, measurements as a function of concentration were also carried out. Localization of fluorescent complexes in tubuli is indicated, as assessed in the experiments shown in Figure 3.

Figure 5 Activity of nuclear -galactosidase in transfected kidneys. (a-c) Sections from a rat left kidney, injected with the nuclear -galactosidase plasmid pCI-galNuc complexed with 0.4 N/P eq of FITC-pLys and with 10 N/P eq of PEI, were prepared 2 days after transfection, fixed and incubated with X-gal. Nuclear staining, indicating specific expression of the transgene, is visible in tubular cells (arrows). No nuclear -galactosidase staining was observed in left kidneys transfected with an irrelevant plasmid (d and e) or in right kidneys of animals whose right kidney had been transfected with the nuclear -galactosidase (f). Bars correspond to 100 m in panels a and d and to 10 m in panels b, c, e and f. Plasmid pCI-galNuc, carrying the LacZ gene from E. coli equipped with a nuclear localization signal under the control of the CMV promoter, was constructed by insertion of the 3.6 kb nuclear LacZ cDNA fragment flanked by KpnI and blunted HindIII sites from pnLacZ (a kind gift of Dr Lawrence Wrabetz, San Raffaele Scientific Institute, Milan) into the KpnI and SmaI sites of plasmid pCISfiT.²¹ One hundred micrograms per animal of the nuclear -galactosidase plasmid were complexed with 0.4 N/P eq of FITC-pLys and with 10 N/P eq of PEI and injected into the left renal artery, as described in the legend to Figure 2. Two days after transfection, kidneys were fixed and 5 mm thick cryosections were prepared as described in the legend to Figure 3, rinsed in 150 mM NaCl and incubated with 200 M chloroquine in PBS at 37°C for 1 h, rinsed in PBS and incubated in 10 mM K₄Fe(CN)₆, 10 mM K₃Fe(CN)₆, 1 mg/ml 5-bromo-4-chloro-3-indolyl -d-galactopyranoside (X-gal), 0.05% sodium deoxycholate monohydrate, 0.05% Nonidet P-40, 5 mM MgCl₂, 0.01 M EGTA in 10 mM phosphate buffer, pH 7.8 at 37°C overnight. After a washing step in PBS, sections were counterstained with haematoxylin and observed at the light microscope (Axioplan, Zeiss).