Gene Delivery in salivary glands: from the bench to the clinic

Abstract

In vivo gene delivery has long been seen as providing opportunities for the development of novel treatments for disorders refractory to existing therapies. Over the last two decades, salivary glands have proven to be a useful, if somewhat unconventional, target tissue for studying several potential clinical applications of therapeutic gene delivery. Herein, we follow the progress, address some problems and assess the outlook for clinical applications of salivary gland gene delivery. Our experience with these tissues provides a roadmap for the process of moving an idea from the laboratory bench to patients.

1. Introduction

The delivery of genes as a therapy for clinical disorders began in earnest in the 1980s [e.g., ^1,2]. Initially, and for many years thereafter, investigations of clinical applications of gene delivery focused on inherited disorders and on malignancies refractory to conventional treatment. However, in the 1990s, many researchers began to explore applying gene delivery to a wide range of conditions, including acquired disorders, for which no suitable conventional therapy then existed. Among the tissues examined at that time for potential clinical applications were salivary glands [e.g., ^3–7].

Although studies with salivary glands are relatively infrequent in the modern translational research literature, these tissues have been the subjects of classic studies in experimental medicine [8,9]. Thus, it is perhaps not surprising that as a result of recent research, salivary glands appear also to have many advantages for clinical gene delivery (see Table 1). Most importantly, salivary glands are (i) easy to access in a relatively non-invasive manner; (ii) well-encapsulated in humans, which limits vector spreading; (iii) confined targets, so that the concentration of vector delivered is not diluted by other body fluids; (iv) easy tissues in which to assess several important physiological processes; and (v) not necessary for life, if a severe untoward complication were to develop. Furthermore, not only are the two major clinical disorders affecting salivary glands (radiation damage subsequent to treatment of head and neck cancers; Sjögren’s syndrome [10,11]) lacking adequate conventional therapy, but also salivary glands are potentially useful for treating monoendocrinopathies that currently do not have a suitable therapy [12,13].

Table 1.

Advantages of salivary glands for clinical gene delivery ¹

The two major disorders affecting salivary glands (radiation damage; Sjögren’s syndrome) lack adequate conventional therapy There is an easy access of a gene transfer vector to almost all epithelial cells in a gland via intraoral cannulation of the main excretory ducts and infusion There is a readily defined fluid volume that can be infused into each gland The volume of infusate fluid is not further diluted following vector delivery The luminal membrane of almost all epithelial cells in the gland is a potential target for infused vectors Salivary epithelial cells divide slowly making a relatively stable target population for non-integrating gene transfer vectors It is easy to measure neurotansmitter coupled secretory responses in salivary glands (i.e., the gustatory stimulation of salivation) Salivary epithelial cells can produce significant levels of transgenic proteins for export in both exocrine and endocrine directions Human salivary glands are well-encapsulated minimizing the potential for vector spread beyond the targeted gland If a severe and life-threatening adverse event were to occur, a single salivary gland is not essential for life and could be readily removed

¹Modified from Baum et al, Trends in Molecular Medicine, 2004 [83]

The latter application has not yet reached the clinic and, thus, is not relevant to the primary focus of this review. However, it helps to illustrate some of the existing difficulties with more general applications of salivary gene transfer. Although exocrine glands, for more than 50 years numerous reports have shown that salivary glands also are able to secrete in an endocrine manner [e.g., ^14–19]. Typically, salivary proteins are secreted via two principal pathways, although additional, minor, subgroups exist: the major regulated pathway and the constitutive pathway [20–22]. Regulated pathway proteins are first stored in secretory granules, where they await an external stimulus for secretion. Thereafter, the cargo proteins in granules are released into the forming saliva (unidirectional secretion). In contrast, constitutive pathway proteins are secreted continuously and non-directionally, roughly at the rate of their translation. Proteins secreted via the constitutive pathway likely account for endocrine secretion from salivary glands.

Over the last ~15 years we have studied the concept of employing salivary glands as a surrogate endocrine organ using several model human proteins, including α-1-antitrypsin (α1AT), erythropoietin (Epo), growth hormone (GH) and parathyroid hormone (PTH) [4,7,12, 23,24]. While results with α1AT and GH have been relatively straightforward, studies with Epo and PTH were not predictable, with differences in their secretory behavior occurring between certain species and gland types [e.g., ^12,23–27]. Unfortunately, little is known about the mechanisms responsible for the sorting of secretory proteins in polarized epithelial cells. As a result, our aggregate studies highlight the need to understand basic cellular mechanisms, in this case secretory protein sorting, before salivary gene transfer for treating monoendocrinopathies routinely can be exploited.

Research on gene delivery to salivary glands, using intraoral, retroductal cannulation, has been performed by at least 16 separate research groups worldwide (Table 2). Salivary gene delivery has been shown to be useful or potentially useful for the repair [28,29] and prevention [30–32] of radiation damage, the treatment of Sjögren’s syndrome [33–35], and gene therapeutics (pharmacological applications) directed at both upper gastrointestinal tract [5,36] and systemic [6,7,23,37] conditions. Additionally, salivary gland gene delivery has been useful for asking biological questions in vivo [38–41], as well as for creating novel models of disease [42,43]. While most of these studies have been performed in rodent models (mice, rats), several studies in large animal models have been reported (miniature pigs, rhesus macaques [25,27,29,44,45]), and a human clinical trial employing gene delivery to parotid glands is ongoing (see below; [46]). Notably, as a result of detecting replication competent adenovirus (Ad) in the saliva of one treated patient in this clinical trial, the notion that vector spread beyond a targeted gland is limited by the gland capsule, has been confirmed [47]. The patient involved apparently had a latent serotype 5 Ad (Ad5) infection in the targeted gland, which was activated following vector delivery. The consequently amplified vector and wild type Ad5 were only found in parotid saliva from the targeted gland, with neither ever detected, using sensitive real time-quantitative PCR assays, in the patient’s serum [47].

Table 2.

Research groups that have shown applicability of salivary gland gene transfer ¹

Radiation damage repair Baum et al, NIDCR, NIH, USA [28,29,71] Wang et al, Capital Medical University, PR China [29,71]

Prevention of radiation damage Baum et al, NIDCR, NIH, USA [31] Greenberger et al, University of Pittsburgh, USA [30] Sunvala-Dossabhoy et al, Louisiana State University, USA [32]

Treatment of Sjogren’s Syndrome Chiorini et al, NIDCR, NIH, USA [33] Tak et al, University of Amsterdam, The Netherlands [34,35]

Gene Therapeutics (exocrine secretion) Baum et al, NIDCR, NIH, USA [5] Huang et al, University of California-Los Angeles, USA [36]

Gene Therapeutics (endocrine secretion Baum et al, NIDCR, NIH, USA [12,23] Passineau et al, Allegheny-Singer Research Institute, Pittsburgh, USA [37] Rothman et al, University of California San Francisco, USA [6]

Physiology and Pathophysiology Ambudkar et al, NIDCR, NIH, USA [38] Baum et al, NIDCR, NIH, USA [4,84] Kawaguchi et al, Tokyo Dental College, Japan [41] Kulkarni et al, NIDCR, NIH, USA [43] Weigert et al, NIDCR, NIH, USA [40]

General gene transfer Barka et al, City University of New York, USA [85] Bennett et al, Genteric, Inc, USA [86] Chiorini et al, NIDCR, NIH, USA [39] Palmon et al, Hebrew University, Israel [87] Passineau et al, Allegheny-Singer Research Institute, Pittsburgh, USA [88]

¹Only senior authors are mentioned for space considerations, with our apologies to the other authors. See [References] for the appropriate citations.

It is the purpose of this review to use some of our group’s experiences to illustrate the process of translating biological science advances from the bench into the clinic.

2. Biotechnology

2.1. Overview of the process

The steps required for moving all potential clinical treatments from the bench to the clinic are multiple, and vary somewhat depending upon the nature of the treatment being considered, e.g., for a gene, a recombinant protein or a traditional type of pharmaceutical. Table 3 provides a fairly detailed listing of the key steps that were involved in moving a gene delivery strategy for repairing radiation damage to salivary glands from a general idea into a clinical trial now being conducted at the US National Institutes of Health-Clinical Research Center ([46]; clinical protocol 06-D-0206; see section 3, below). This study involves using a recombinant Ad5 vector encoding the archetypal water channel protein, human aquaporin-1 (hAQP1 [48]; AdhAQP1) for the treatment of a single parotid gland in patients who have received radiation therapy for the treatment of a squamous cell carcinoma in the head and neck region. To go from the first step (identification of the problem) to the last step (initiating patient enrollment in this study) took us 16 years. This is a fairly long time, albeit not unusual, and required considerable and stable institutional support.

Table 3.

Steps in taking a potential gene therapy from the bench to the clinic ¹

Identification of a clinical problem without conventional therapy Development of the therapeutic idea using gene transfer Understanding the biology of the intended target tissue Assessment of the risk/benefit ratio for using viral and non-viral vectors Decision on the vector to be used (herein, a viral vector) Understanding the biology of the vector to be used Determining the availability of suitable in vitro and in vivo (small and large animal) models for testing the idea Construction of the vector to be used Functional testing of the vector in vitro Efficacy studies with the vector in a small animal disease model Efficacy studies with the vector in a large animal disease model Conducting a toxicology and biodistribution study with the vector in small animals (GLP level2) Development of a clinical protocol Required reviews and approval of the clinical protocol3 Production of a clinical grade gene transfer vector (GMP4) Establishment of the infrastructure required to support the study Patient enrollment

¹Modified from Baum et al, Oral Oncology, 2010 [46]

²GLP, Good Laboratory Practice as defined by the US Food and Drug Administration

³for the AdhAQP1 clinical study discussed herein, five separate reviews were required: NIDCR Institutional Review Board, NIH Biosafety Committee, US Recombinant DNA Advisory Committee, US Food and Drug Administration, and the study’s Data Safety and Monitoring Board

⁴GMP, Good Manufacturing Practice as defined by the US Food and Drug Administration

At the time the idea of a potential gene therapy for salivary gland radiation damage was crystallized, we had none of the knowledge, skills or experience required to work with recombinant viral vectors, though we were well versed in salivary gland biology. It took three years for us simply to demonstrate the feasibility of salivary gland gene transfer using an Ad5 vector and intraductal cannulation [3], and an additional three years to show efficacy in the first animal model (rat) tested [28]. For developing clinical applications of gene delivery, and other types of biological therapies, it is essential to show scaling to a large animal model [49]. This step was quite time-consuming (8 years), as we initially tried to conduct a study in rhesus macaques [50], but found it much too expensive to test appropriate numbers of animals. We then switched to a miniature pig model [44,51], and eventually showed efficacy in a study using a fairly large number of animals (n=18, [29]).

The actual process of protocol development and approval took about 3 years. This included reviews by five separate groups: the NIDCR Institutional Review Board (IRB), the US Recombinant DNA Advisory Committee (RAC), the NIH sBiosafety Committee, the US Food and Drug Administration (FDA), and the study’s Data Safety and Monitoring Board (DSMB). We were quite fortunate to have used two consultants to help with shaping the final clinical protocol and the Investigational New Drug (IND) application to the FDA. Both individuals, one a PhD molecular biologist and the other a physician, had considerable prior work experience at the FDA and, thus, were able to provide us with both tangible guidance in required procedures, as well as a perspective on the entire process. As a result of their guidance, the protocol and IND applications were well prepared, and the review and approval process proceeded in a timely manner. The time from the initial submission to the IRB, until the final approval of the DSMB permitting us to commence enrolling patients, was ~18 months, including a public review of the protocol by the RAC. Each of the five groups involved in the approval process was thorough and made inquiries consistent with their responsibilities and the nature of the protocol. Given that this protocol was the first gene therapy proposed for a salivary gland, designed to treat a condition affecting quality of life, albeit one without any existing treatment, to us the entire process seemed fair, constructive and appropriate.

The production of the clinical grade AdhAQP1 vector and establishment of the required infrastructure, most notably developing an electronic database containing all required case report forms and creating a study monitoring system, was begun during the latter stages of the review and approval process and was completed ~10 months after all approvals were obtained. Although our institute had minimal infrastructure for conducting a study of this type, it provided us with an experienced Contract Research Organization (CRO) to establish what was needed. Similarly to our positive experience with the protocol and IND application consultants, we benefited greatly from working with a CRO that knew what was required to ensure that the AdhAQP1 study was conducted safely and efficiently. In hindsight, we probably should have begun the infrastructure development process sooner, as absent proper infrastructure patient enrollment was delayed significantly. However, we chose to proceed cautiously and waited until late in the approval sequence before commencing.

2.2. Method of gene delivery to salivary glands

Clinical gene delivery to major salivary glands (parotid and submandibular) can be considered a minimally invasive procedure. The technique, which requires no local anesthesia, is based on cannulation of Stensen’s (parotid) or Wharton’s (submandibular) ducts, which are readily accessible in the mouth. In fact, such cannulation is routinely used in taking contrast x-rays (sialography) of the glands [52]. During the procedure, the duct orifice is first identified (surgical magnification loops are useful), and then probed to allow access and cannulation with a blunt-ended catheter (Figure 1). Introduction of the vector (or contrast agent in the case of sialography) is performed typically by hand injection, but could also be performed using a continuous-infusion pressure-monitored technique. In humans, the hydrostatic pressure allows the infusion of an average 0.5–1.5 ml volume (determined exactly for each patient), and prevents overfilling of the gland and reduces discomfort [53].

Fig. 1.

Administration of gene transfer vectors to an animal and a patient. (a) Photograph of rat with both submandibular glands cannulated through the orifices of Wharton’s duct in the floor of the mouth. Each cannula is connected to a syringe (bottom right and left of the figure), through which a gene transfer vector is being infused. Animals undergoing salivary gland gene transfer are anesthetized only for restraint. (b) Photograph of a patient in NIH clinical protocol 06-D-0206 who has the left parotid gland cannulated through Stensen’s duct in the buccal mucosa and is receiving the AdhAQP1 vector through a syringe attached to the cannula.

Gene delivery to salivary glands of animals is performed in a similar fashion, although anesthesia is used only for restraint. The cannulation of salivary ducts in rodents requires the use of a stereomicroscope and small catheters. In fact, we routinely prepare catheters by stretching polyethylene PE-10 tubing over an open flame, thinning them to the appropriate diameter. To prevent back-flow of the infusate following its administration in animals, we use cyanoacrylate to temporarily adhere the catheter to the duct opening (Figure 1A; rat submandibular gland is cannulated). Vector volume is important for attaining maximal transgene expression. The optimal volume varies by species and is 50 and 200 μl for mouse and rat submandibular glands, respectively [54,55], 500 μl for the parotid glands of rhesus macaques [45], and 4000 μl for a miniature pig parotid gland [44].

2.3. Principal vectors used for salivary gene transfer

Transfer of genes into cells can be accomplished using viral and non-viral vectors. Currently, viral vectors provide the most efficient means for gene transfer, but they raise safety concerns including, in particular, the risk of insertional mutagenesis and the possibility that their manufacture or use will generate replication-competent viruses. Additionally, viral vectors can elicit immune responses, which can be marked and prevent the repeated use of the product [56–58]. In contrast, the delivery of DNA via non-viral vectors poses less of a safety risk, but is inefficient in transducing mammalian cells. Methods for non-viral gene delivery use physical and chemical approaches to improve efficacy and cell specificity. Direct injection, a gene gun, electroporation, sonoporation (microbubbles) and laser irradiation have all been studied as physical means to improve gene transfer [59–63]. Chemical approaches aim to enhance endosomal escape, intracellular trafficking and nuclear import of the transferred gene and use cationic liposomes or polymers most frequently [64]. For salivary glands, non-viral vectors are seldom employed, while Ad5 or serotype-2 adeno-associated viral (AAV2) vectors are most often used.

By 2009, over 350 protocols (including NIH # 06-D-0206; see below) using Ad5 vectors have been approved for clinical trials. Ad5 vectors are also extremely useful in proving a concept for potential clinical application. They readily transduce salivary gland epithelial cells in animals (mice, rats, miniature pigs, and non-human primates) and produce a high level of expression of the delivered gene, albeit transiently (typically 7–14 days [65]), because of a considerable immune response. In comparison, AAV2 vectors induce substantially less host immune reactivity and thus AAV2-mediated transgene expression lasts much longer [45]. Although AAV2 vectors are more difficult to construct, in great part because their biology is less understood than that of Ad5, AAV2 vectors are quite useful in studies that require long-term expression. Importantly, wild type AAV2 is not associated with any known pathology in humans.

3. AdhAQP1 to correct radiation-induced parotid hypofunction

Salivary glands are composed of two distinct major cell types: acinar cells and duct cells (Figure 2). Acinar cells are water-permeable and NaCl secreting, i.e., an acinus is a secretory epithelium. Acinar cells are considered to be the only site of fluid movement in these glands, and generate an isotonic, so-called “primary” fluid [66,67]. Duct cells are essentially water-impermeable and reabsorb considerable NaCl making the final secreted saliva markedly hypotonic, i.e., the salivary duct system behaves as an absorptive epithelium. Radiation used in the treatment of head and neck cancers leads to the loss of acinar cells in the radiation field, which is likely a result of several targeted insults [10]. Consequently, damaged salivary glands in surviving head and neck cancer patients have predominantly duct cells remaining, i.e., unable to secrete fluid, and the patients experience significant salivary hypofunction, often termed xerostomia (dry mouth). With a five-year survival rate of ~60%, such patients (Radiation Therapy Oncology Group categories 2 and 3, [68]) suffer from dysphagia, frequent oral infections, poor oral mucosal wound healing and a marked reduction in quality of life [69,70], and there is no adequate conventional therapy available.

Fig. 2.

Schematic depiction of the gene transfer strategy being employed in the AdhAQP1 clinical trial (NIH clinical protocol 06-D-0206). See text for details. This figure was originally published in Oral Oncology [46] and is reprinted with permission.

As depicted in Figure 2, we posited that, in the absence of significant acinar cell primary fluid formation, duct cells were capable of generating a lumen>interstium osmotic gradient [28,46]. This conjecture, based on limited published data obtained using rodent models, assumed the formation of a KHCO₃⁻-driven osmotic gradient [discussed in ^28,29,71], but the exact mechanism remains unproven. However, in order to achieve actual fluid secretion, whatever the nature of the osmotic gradient, water would be unable to follow since duct cells lacked facilitated water permeability pathways (water channels) in their plasma membranes. Serendipitously, at the time we were developing our ideas, Preston and Agre published a report describing the isolation of the cDNA for hAQP1 [48]. Thus, we hypothesized that delivering the hAQP1 cDNA to duct cells in radiation-damaged salivary glands could lead to fluid secretion by providing facilitated water permeability pathways in these cells. This hypothesis was tested initially using the AdhAQP1 vector to transduce cells in irradiated rat submandibular glands [28]. Under the optimal conditions employed, this maneuver resulted in near normal salivary flow rates in the irradiated animals, while a control Ad5 vector was without benefit [28].

Thereafter, as noted above, we decided to test the same hypothesis in a large animal model, targeting the parotid glands of rhesus macaques. This model was chosen because of (i) its obviously close relationship to humans and (ii) past experiences studying parotid saliva secretion in this species [e.g., ^72,73]. We were able to obtain five, male, out-bred rhesus macaques (~8–10 kg each) for study and decided to use two animals in each of two AdhAQP1 dosage groups with one animal to receive an Ad5 control vector [50]. In retrospect, this was an ill-advised experimental design, as the small number of animals could not accommodate normal variations in salivary secretion, as well as variability in radiation damage, and the final results were equivocal [50].

We decided, next, to repeat the AdhAQP1 rat study, and obtained results generally comparable to those of our original study [74]. With that impetus, we then re-explored possible large animal models to use. An important consideration was that the model would be more affordable and less labor intensive for us to manage. We decided to use miniature pigs, collaborating with a research group in China. These animals are much larger than macaques (~25–30 kg), markedly less expensive, and it had been shown previously by our collaborators that their parotid glands were in many ways similar to those of humans [51]. Using an Ad5 vector encoding the luciferase reporter gene, we demonstrated good dose scaling between mice and miniature pigs [44], as well as significantly greater radiation sensitivity than that of rats [75,76]. Accordingly, we then tested our original hypothesis in miniature pigs whose parotid glands had been subjected to radiation. We obtained similar results to those found in the rat studies (Figure 3, [29]), using three separate cohorts of miniature pigs and, in fact, much lower doses of the AdhAQP1 vector.

Fig. 3.

Results from the key preclinical efficacy test of the AdhAQP1 gene transfer strategy shown in Figure 2. The parotid glands of miniature pigs were irradiated at time zero and after 16 weeks parotid salivary flow (normalized to 100%) was reduced by ~80 %. At week 17, animals were given the AdhAQP1 vector, or a control vector encoding luciferase, in the irradiated glands. Parotid salivary flow increased promptly thereafter in the AdhAQP1-treated glands. See text for additional details. This figure was originally published in Molecular Therapy [29] and is reprinted with permission.

All of our studies in rats, rhesus macaques and miniature pigs also had suggested that delivery of an Ad5 vector to the salivary glands was generally safe. However, in order to consider a possible clinical trial, we needed to conduct detailed safety studies, monitoring potential toxicological responses, as well as vector bio-distribution, after salivary gland delivery. As indicated in Table 3, we also had to conduct this study under the Good Laboratory Practice (GLP) guidelines established by the US FDA. These studies, using 100 male and 100 female rats, were conducted in collaboration with colleagues at the US National Toxicology Program [77]. In addition to a control group, administered the buffer diluent, three doses of AdhAQP1, which we thought would span the dose range expected for use in a clinical trial, were tested. The results were similar to those observed in our non-GLP studies, i.e., delivery of AdhAQP1 to a salivary gland was essentially safe and vector was primarily contained within the targeted gland [77].

Thereafter, based on the small and large animal efficacy studies, and the rat safety study, we began to develop a clinical protocol, which received all five required approvals, and concurrently arranged for the clinical grade vector to be produced by the Belfer Gene Therapy Core Facility at Weill Medical College of Cornell University (New York, NY). More detailed information about the approved protocol, “Open-label, dose-escalation study evaluating the safety of a single administration of an adenoviral vector encoding human aquaporin-1 to one parotid salivary gland in individuals with irradiation-induced parotid salivary hypofunction” (NIH # 06-D-0206), can be found at the Clinicaltrials.gov web site (http://www.clinicaltrials.gov/ct/show/NCT00372320?order=) and at the study web site (http://www.drymouthstudy.com/), as well as in a previous, brief review [46]. Fifteen patients were approved for treatment under this protocol, in five dosage groups, from 4.8×10⁷ – 3.5×10¹⁰ vector particles/gland. Eleven patients have been treated as of this date (June 21, 2011), and all patients have tolerated the vector and study procedures well. Formal analysis of the efficacy data has not yet been performed, and reports of the safety and efficacy data will be submitted separately for publication following study completion.

4. Conclusions

There is now a considerable body of published literature to demonstrate the feasibility of using gene delivery to salivary glands for clinical applications. While to date, there is only a single approved clinical trial employing salivary gland gene delivery, already some key advantages of targeting this tissue have been confirmed in patients. As in all animal models, not surprisingly, there is an easy access and delivery of patient-specific volumes of vector to the targeted glands, and vector also appears to be well contained within the gland.

The rate of progress in translating basic and pre-clinical research into clinical applications for gene delivery to salivary glands, however, has been modest. In part, this is certainly a reflection of progress generally in the gene therapy field [78], as well as the general nature of conducting bench to clinic studies [79,80]. Additionally, although many research groups have demonstrated feasibility (Table 2), the number of investigators actively pursuing clinical applications of salivary gland gene delivery is relatively small; as far as we know, only six of the research groups listed in Table 2. This relative paucity of investigators obviously also affects rates of progress. Furthermore, this already limited number is coupled with a relatively small number of investigators who study fundamental questions about the biology of salivary glands, also hindering bench to clinic progress with this tissue. An excellent example of the latter, discussed above, relates to the problem of understanding signals that differentiate secretory protein sorting to the regulated versus constitutive pathways, i.e., exocrine versus endocrine secretion. An understanding of this fundamental biological issue is needed before full advantage of using salivary glands as a target tissue for systemic gene therapeutics can be achieved [81].

Nonetheless, there are real and valuable translational possibilities available through salivary gene delivery. The process of moving science between the bench and the clinic is complex and generally slow [82], and is filled with many “bumps in the road” [82]. However, it is achievable, as we hope has been demonstrated herein.

Highlights.

• This review offers a roadmap of bench to clinic research using salivary gland gene transfer applications as an example

• Advantages of salivary glands as gene transfer target sites are described

• A key clinical application (repair of radiation-damaged glands) is emphasized

• The opportunities and difficulties associated with bench to clinic research are considered