In Vivo Protein Transduction: Delivery of a Biologically Active Protein into the Mouse

See allHide authors and affiliations

Science  03 Sep 1999:
Vol. 285, Issue 5433, pp. 1569-1572
DOI: 10.1126/science.285.5433.1569


Delivery of therapeutic proteins into tissues and across the blood-brain barrier is severely limited by the size and biochemical properties of the proteins. Here it is shown that intraperitoneal injection of the 120-kilodalton β-galactosidase protein, fused to the protein transduction domain from the human immunodeficiency virus TAT protein, results in delivery of the biologically active fusion protein to all tissues in mice, including the brain. These results open new possibilities for direct delivery of proteins into patients in the context of protein therapy, as well as for epigenetic experimentation with model organisms.

Currently, efficient delivery of therapeutic compounds, peptidyl mimetics, and proteins into cells in vivo can be achieved only when the molecules are small—typically less than 600 daltons (1). Delivery of bioactive peptides across the blood-brain barrier, for example, is generally restricted to small (six amino acids or less), highly lipophilic peptides (1). Gene therapy (2) is one promising method for circumventing this problem, but conditions for high-efficiency targeting and long-term protein expression have yet to be discovered.

We have focused on an alternative approach of “protein transduction,” or protein therapy, to address this problem. In this method (3), full-length fusion proteins are generated that contain an NH2-terminal 11–amino acid protein transduction domain (PTD) from the human immunodeficiency virus (HIV) TAT protein [first identified in 1988 (4)]. These proteins are then purified under denaturing conditions (3). Protein transduction occurs in a rapid, concentration-dependent fashion that appears to be independent of receptors and transporters (5) and instead is thought to target the lipid bilayer component of the cell membrane. Thus, in principle, all mammalian cell types should be susceptible to protein transduction, and indeed we have used this technology to transduce over 50 proteins ranging in size from 15 to 120 kD into a wide variety of human and murine cell types in vitro (3,6).

To determine whether this method could be used to deliver peptides in vivo, we synthesized a 15-oligomer peptide containing the 11–amino acid TAT PTD, preceded by an NH2-terminal fluorescein isothiocyanate (FITC)-Gly-Gly-Gly-Gly motif that rapidly transduced into ∼100% of cultured cells (7). We next injected C57BL/6 mice intraperitoneally with 1.7 nmol of the TAT-FITC peptide or with control free FITC and monitored the appearance of fluorescent cells (8). Flow-activated cell sorting (FACS) analysis of whole blood isolated 20 min after intraperitoneal (ip) injection with TAT-FITC peptide revealed a strong fluorescence signal in ∼100% of blood cells (Fig. 1A, left). Blood cells from mice injected with control free FITC showed a small constant increase in background fluorescence (9) that was likely due to uptake of FITC from the peritoneum by the lymphatic system. We also analyzed splenic cells by performing a splenectomy 20 min after ip injection of the mice. FACS analysis revealed transduction of TAT-FITC peptide into all splenic cells, including T cells, B cells, and macrophages (Fig. 1A, right). Control ip injections of equimolar amounts of free FITC showed only a minor increase in fluorescence above background levels. Thus, the injected TAT PTD peptide was rapidly transduced into all blood and splenic cells.

Figure 1

Transduction of TAT-FITC peptide into mice. (A) Flow cytometry of whole blood cells (left) and splenocytes (right) isolated from mice 20 min after ip injection of TAT-FITC peptide or control free FITC. (B) Fluorescence confocal microscopy of hemispheric sagittal brain sections (top) and skeletal muscle tissue sections (bottom) isolated from mice 20 min after ip injection as in (A). Scale bars, 100 μm.

We next studied the uptake of the TAT peptide into brain tissue and skeletal muscle. Tissues were dissected from mice 20 min after ip injection with TAT-FITC peptide (8), and cryostat sections were prepared. Fluorescence confocal microscopy analysis of 10-μm hemispheric sagittal brain sections revealed a strong signal in all areas of the brain from TAT-FITC peptide-injected mice, whereas the signal in control FITC–injected mice remained at background levels (Fig. 1B, top). Fluorescence photobleaching was observed when TAT-FITC peptide sections were subjected to prolonged excitation, providing further evidence that the TAT-FITC peptide was in the brain section. Skeletal (quadriceps) muscle also showed a significant fluorescence signal from TAT-FITC–injected mice as compared to control FITC–injected mice (Fig. 1B, bottom).

To determine whether a large biologically active protein could be successfully transduced in vivo, we used the 116-kD β-galactosidase (β-Gal) protein (10). An NH2-terminal TAT–β-Gal fusion protein (120 kD) was generated (11), as was a control β-Gal fusion protein missing the 11–amino acid TAT PTD but retaining the rest of the NH2-terminal leader (119 kD) (Fig. 2A). Fluorescence confocal microscopy of cultured cells treated with FITC-labeled TAT–β-Gal (12) revealed that the protein was inside the cells, whereas control β-Gal–FITC was not detectable, and control free FITC was bound to the cellular membrane (Fig. 2B). Immunoblot analysis revealed that the TAT–β-Gal was rapidly transduced into cultured cells, reaching near maximum intracellular concentrations in less than 15 min, whereas the control β-Gal protein did not transduce into cells even after 2 hours (Fig. 2C). Unexpectedly, the enzymatic activity of TAT–β-Gal peaked about 2 hours later than did the intracellular concentration (13) (Fig. 2C). This lag may reflect a slow posttransduction refolding rate of the protein by intracellular chaperones such as HSP90 (14).

Figure 2

Transduction of TAT–β-Gal into cultured cells. (A) Diagram of TAT–β-Gal and control β-Gal fusions. (B) Fluorescence (top) and light (bottom) confocal microscopy showing that TAT–β-Gal–FITC entered Jurkat T cells, whereas β-Gal–FITC did not. (C) Concentration (left axis) and enzymatic activity (right axis) of HepG2 cells treated with TAT–β-Gal (open squares) or β-Gal (solid circles). β-Gal concentration was determined by immunoblot analysis. (D) Flow cytometry of whole blood cells (left) and splenocytes (right) performed at the indicated times after ip injection of mice with TAT–β-Gal–FITC or control β-Gal–FITC.

We next intraperitoneally injected FITC-labeled TAT–β-Gal and control β-Gal proteins into mice. Flow cytometry of blood and splenic cells isolated 30 and 120 min, respectively, after ip injection demonstrated the presence of TAT–β-Gal–FITC in all blood and splenic cells (Fig. 2D). In contrast, the control β-Gal–FITC was not detectable (Fig. 2D). Similar results were obtained in experiments with two other FITC-labeled TAT fusion proteins: TAT-Cdk2-DN [human Cdk2 (cyclin-dependent kinase-2) dominant negative] (36 kD) (3) and TAT-CAK1 [yeast CAK1 (Cdk-activating kinase-1)] (47 kD) (15, 16).

The mice were then analyzed for β-Gal enzyme activity. Tissue samples from the liver, kidney, heart muscle, lung, and spleen were isolated at 4 and 8 hours after ip injection, sectioned (in 10- and 50-μm sections), and assayed by X-Gal staining (8) (Fig. 3). Liver, kidney, and lung tissues from TAT–β-Gal–injected mice showed strong and uniform β-Gal activity across the tissue sections at 4 and 8 hours after ip injection. The heart samples also showed strong β-Gal activity throughout the muscle fibers. Sections from control β-Gal–injected mice showed either no staining or sporadic weak staining that was likely due to lymphatic uptake of control β-Gal from the peritoneum. The control β-Gal–treated kidney showed weak staining, presumably reflecting clearance from the bloodstream. The spleen showed strong β-Gal activity in the red pulp areas and much weaker activity in the white pulp areas (Fig. 3), which are principally composed of T and B cells. This is at odds with the FACS transduction analysis (Fig. 2D); however, T and B cells are known to contain a β-Gal–inhibitory activity (17).

Figure 3

Transduction of TAT–β-Gal into mice. Analysis of β-Gal enzymatic activity assessed by X-Gal staining in the liver, kidney, lung, heart muscle, and spleen is shown. Samples were analyzed 4 hours after ip injection of proteins, except for kidney tissue, which was analyzed 8 hours after injection. There is weak β-Gal activity in the white pulp areas of the spleen. Sections were developed for 4 hours. Scale bars, 100 μm.

To determine whether the protein crossed the blood-brain barrier, we performed X-Gal staining on mid-hemispheric sagittal brain sections from mice at various times after ip injection of TAT–β-Gal or control β-Gal (Fig. 4). Brain sections from mice analyzed 2 hours after ip injection with TAT–β-Gal showed strong activity that was principally localized around blood vessels, with minimal activity being present in the surrounding parenchyma (Fig. 4A). However, by 4 hours after ip injection, all regions of the brain showed strong β-Gal activity. In contrast, mice injected with control β-Gal showed no β-Gal activity in the brain at 2 hours (15), 4 hours (Fig. 4A), or 8 hours after injection (15). By 8 hours after ip injection of TAT–β-Gal, enzyme activity resided primarily in cell bodies throughout all regions of the brain, and to a lesser extent in the surrounding white matter. Because cell bodies are mainly composed of nuclei, this raises the possibility that the protein enters the nucleus, perhaps via the embedded nuclear localization signal in the TAT PTD (18) (Fig. 4A). The blood-brain barrier remained intact in TAT–β-Gal–treated mice, as measured by the absence of extravasated coinjected Evan's blue albumin complexes in brain sections (19) (Fig. 4B). In addition, a low-magnification coronal brain section revealed β-Gal activity in cell bodies throughout the brain of TAT–β-Gal–treated mice 8 hours after injection (Fig. 4C).

Figure 4

Transduction of TAT–β-Gal across the blood-brain barrier. (A) β-Gal activity (X-Gal staining) in hemispheric sagittal brain sections from mice injected intraperitoneally with TAT–β-Gal or control β-Gal; sections were made after the indicated times. Eight hours after ip injection, TAT–β-Gal localized primarily to the nuclei of cell bodies (arrows) throughout the brain section. Sections were developed in X-Gal for 16 hours. Scale bars, 100 μm. (B) Extravasated Evan's blue dye was detected adjacent to blood vessels throughout the brain sections (50 μm) from mice treated with the protamine positive control (top) but was not detected in sections from TAT–β-Gal–treated mice (bottom). Scale bars, 100 μm. (C) β-Gal activity in a low-magnification coronal brain section from mice 8 hours after ip injection with control β-Gal (left) or TAT–β-Gal (right), showing activity through-out the brain. Arrows indicate the region of endogenous β-Gal activity. Scale bars, 500 μm.

A previous attempt to transduce β-Gal chemically cross-linked to the TAT PTD into mice resulted in sporadic and weak β-Gal activity in a limited number of tissues, with no activity detected in the kidney or brain (20). The increased transduction potential reported here likely reflects the in-frame fusion and purification strategy (3). Furthermore, we have engineered a series of artificial PTDs that show dramatically enhanced transduction potential in cultured cells (21).

The transduction of peptides and proteins into mice lays the groundwork for future epigenetic complementation experiments in model organisms and for the eventual transduction of therapeutic proteins into patients in the form of protein therapy. Therapy with biologically active full-length proteins will allow access to the built-in evolutionary specificity of these proteins for their targets, thereby potentially avoiding the nonspecific effects sometimes seen with small molecule therapies. To this end, we have described an experimental anti-HIV protein therapy based on a transducible HIV protease-activated caspase-3 that selectively induces apoptosis in HIV-infected cells (22). Finally, this methodology opens new possibilities for the development of vaccines and protein therapies for cancer and infectious diseases. Potential immune responses and toxicity associated with long-term transduction of proteins in vivo are important issues that remain to be examined. Along these lines, we note that injection of a mouse with 1 mg of a TAT PTD fusion protein per kilogram of body weight each day for 14 consecutive days produced no signs of gross neurological problems or systemic distress (23).

  • * To whom correspondence should be addressed. E-mail: dowdy{at}


View Abstract

Stay Connected to Science

Navigate This Article