Design and initial characterization of a novel mini-promoter for gene therapies targeting the central nervous system

Abstract

Choosing an optimal combination of adeno-associated virus (AAV) capsid and promoter is key to achieving precise, efficient, and safe gene delivery. However, the limited packaging capacity of AAV can make incorporating large transgenes with strong promoters challenging. To address this, we screened several promoter designs derived from portions of the 5′ untranslated regions (UTRs) of the human ERCC6 and ERCC8 DNA repair genes to generate a novel 126-base pair (bp) mini-promoter (CP040). We evaluated the expression capabilities of CP040 and its components in vitro and demonstrated that intravenous (i.v.) administration of AAVDJ vector harboring enhanced green fluorescent protein (eGFP) driven by the CP040 promoter (AAVDJ-CP040-eGFP) in neonatal mice resulted in strong central nervous system (CNS) expression 6 weeks later. CP040 expression was compared with the chicken beta actin (CBA), human elongation factor 1 alpha (EF1α), and human synapsin 1 (hSYN) promoters following intracerebroventricular (ICV) and intrathecal (IT) AAVDJ administrations. The CP040 mini-promoter yielded comparable expression levels to CBA and hSYN promoters and higher levels than EF1α, with lower off-target expression in liver and kidneys. These initial characterizations support CP040 as a promising candidate to enhance AAV-mediated gene therapy targeting the CNS when cargo sequences cannot fit within AAV size constraints.

Graphical abstract

Pacak and colleagues invented a 126-bp mini-promoter derived from human ERCC6/ERCC8 5′UTRs (CP040) that enables efficient AAV-mediated CNS expression. AAVDJ-CP040-eGFP achieved strong CNS expression that was comparable to CBA and hSYN promoters, outperforming EF1α, while minimizing off-target activity in liver and kidney supporting its utility for size-constrained AAV gene therapies.

Introduction

Adeno-associated virus (AAV) is a top gene delivery vehicle for human therapeutic applications to treat inherited disorders. AAV efficiently delivers genetic material into non-dividing cells and persists as an episome within the nucleus.^1,2 Despite its implementation in several US Food and Drug Administration (US FDA)-approved therapies, limitations related to tissue specificity, packaging capacity, and immunogenicity present barriers to the development of AAV-based therapies for certain larger genes.^3,4,5,6,7

Several strategies can be employed to influence patterns of AAV transduction and transgene expression in vivo, including capsid selection, route of administration, and promoter selection/design. An ideal AAV capsid (serotype) targets desired tissues and de-targets others.^8,9 The ability to avoid transduction of or transgene expression within off-target tissues is arguably as important as providing transgene expression to target tissues. There may be particularly fragile tissues in specific disease populations (such as the liver and kidney for Cockayne syndrome [CS]^10,11,12,13) or tissues such as liver that trigger immune responses following systemic delivery of high AAV doses.^{14,15,16,17,18,19} Certain routes of administration further restrict expression to the desired tissue targets.^8,20 To optimize AAV-mediated targeting of neurological tissue, we recently completed a study that compared two AAV capsid serotypes. Our study found that AAVDJ was superior to AAV9 when assessed for both transduction of desired targets (central nervous system [CNS]) and avoidance of undesired targets (liver and kidney).⁸

The choice of promoter also has a significant impact on patterns of transgene transcription in specific cell types.^{21,22,23,24,25} Promoters are essential elements of gene therapy vector designs that directly influence transcription levels. For AAV-based gene therapies, once AAV vector genomes (vgs) are successfully incorporated into the nuclei of cells, the promoter sequence is required for the efficient recruitment of transcriptional machinery needed for successful transcription of the delivered transgene. Regulatory elements within promoters also influence transgene expression, amplifying or dampening the primary effects of the promoter sequence itself in specific cell types.^26,27,28,29

Many therapeutic AAV constructs employ ubiquitous, virally derived promoters to drive expression; however, this strategy can lead to detrimental off-target side effects as overexpression of some transgenes can be toxic, triggering immune reactions and inflammation.³⁰ Another problem is that host responses to the presence of viral sequences can sometimes lead to transgene silencing through methylation.^31,32 Thus, for both safety and durability reasons, there is a strong rationale to develop and use gene therapy promoters derived from mammalian sequences. One option is to use endogenous promoters known to be functional in the specific cell types of interest. These tend to confer more physiologically accurate levels of expression within cells where they are naturally active drivers of transcription.^33,34,35 One persistent challenge when using AAV and evaluating promoter options is its limited packaging capacity of ∼4,900 bp.³⁶ This often precludes the use of many well-characterized promoters due to the need to also deliver a large transgene (ERCC6 for CS B [CSB]) or genome/prime editing tools (CRISPR-Cas9, etc.). In such situations, the use of compact or miniaturized versions of promoters is needed.

Despite tremendous interest in the field, there is a scarcity of mini-promoters that can efficiently drive the expression of long transgenes. Due to the size of ERCC6 (one of the two classic CS genes) and the packaging constraints of AAV, we needed to find a reliable small promoter (126 bp or shorter) that drives expression in neurological tissues for our CS gene therapy program. CS is a rare, inherited DNA repair disorder that is characterized by an accelerated aging phenotype that affects the entire body and prominent, severe neurodegenerative complications, including cognitive dysfunction, brain atrophy, and dementia.³⁷ We designed and characterized a mini-promoter derived from a combination of two human CS genes (excision repair cross complementation 6 and 8 [ERCC6 and ERCC8]) promoter sequences and transcriptional binding sites designed to bolster transgene expression in the CNS.

While our primary inspiration for designing a mini promoter was due to the vector size constraints limiting our options for promoters in AAV vectors to treat CSB, upon observing our promoter’s activity in vitro and in vivo, we realized that it has the potential for wider translational applications beyond CS. ERCC6 (CSB) and ERCC8 (CSA) are naturally expressed ubiquitously to support DNA repair mechanisms; they demonstrate robust expression throughout the CNS including neurons and glial populations.³⁸ Therefore, our design establishes a natural disease-relevant mini-promoter for gene therapy applications in which the CNS is a primary target. This does not restrict the applicability of the promoters to CS; rather, it strengthens their translational potential across a range of neurological disorders requiring widespread neuronal and glial transduction. Also, to the best of our knowledge, ERCC6- and ERCC8-derived promoter sequences have not yet been explored for their potential in promoter engineering. Thus, this is a novel approach using a combination of promoter elements that are unique in gene therapy settings.

In the current study, we designed several mini-promoter variants, performed a screen to determine which fragment or variant of the mini-promoter was most essential for achieving robust expression, and tested the performance of the selected promoter following intravenous (i.v.) AAVDJ-CP040-eGFP administration in mice. We directly compared our optimal mini-promoter variant with two common and relatively ubiquitous promoters—chicken beta actin (CBA) and human elongation factor 1 alpha (EF1α), and the neuron-specific promoter human synapsin 1 (hSYN) (which are all larger than our limited 126 bp of available promoter space for ERCC6 gene delivery). These comparisons were performed using two different routes of AAVDJ administration—intrathecal (IT) and intracerebroventricular (ICV) in adult wild-type (WT) mice. We assessed vg titer, enhanced green fluorescent protein (eGFP) transcription assays, and immunofluorescence (IF) imaging to quantify the extent of eGFP expression and to compare the performance of these promoters across neurological tissues and cell types.

Here, we present the design, screening, and testing of a promising, small (126-bp) promoter that will be a valuable asset for our gene therapy programs and for other gene therapy applications targeting the CNS.

Results

To design a mini-promoter, we searched for sequences containing high densities of transcription factor-binding sites within the 5′ untranslated regions (UTRs) of the human ERCC8 and ERCC6 genes. Candidate excerpts from the ERCC8 and ERCC6 5′ UTRs were ligated to either side of a Spacer 2 sequence containing additional binding sites, GATA-1, Oct-1, AP-1, c-Jun, and CRE-BP1. To augment promoter activity, a GC-rich Spacer 1 sequence³⁹ was ligated to the 5′ end of the ERCC8 UTR excerpt. Finally, at the 3′ end of the promoter, immediately following the ERCC6 UTR region, a Spacer 3 stabilizer sequence containing RAR-a-, RXR-a-, T3R-, MyoD-, and COUP-binding sites was included. Using ALGGEN’s Prediction of Transcription Factor Binding Sites (PROMO) tool, the mini-promoter sequence was trimmed to maintain essential promoter-binding sites (Figure S1) and reduce the total length to 126 bp. The final version of the full-length mini-promoter (CP040) sequence contained five fragments (Figure 1A). This sequence was custom synthesized in the form of a plasmid and subcloned upstream of the reporter genes contained in the vectors used throughout this study (Table 1).

Figure 1.

CP040 mini-promoter sequence and in vitro characterization

Schematic representation of the full-length sequence containing different fragments (A) 1–5 and (B) variants 1–4. These fragments and variants of the CP040 sequence were cloned between the KpnI and XhoI restriction sites upstream of the firefly luciferase gene in a promoter-less (pGL4.20) vector. The constructs were co-transfected with control (CMV-renilla) plasmid in (C) HEK293 or (D) NPCs. Lysates were prepared, and renilla (control) and firefly (test) luminescence levels were measured and fold change in relative light units calculated compared to empty vector and presented for each fragment and variant in the graphs. The CBA promoter was used as a positive control. Data represent mean ± SEM of three independent replicates (N = 3, ∗∗∗∗p < 0.0001 by ANOVA).

Table 1.

The details of promoter constructs used for dual luciferase assay

pGL4.20 vector	Description	Length (bp)	Cloning sites
Variant 1	Full-length CP040	126	KpnI and XhoI
Variant 2	CP040 without Spacer 1	114	KpnI and XhoI
Variant 3	CP040 without Spacer 3	107	KpnI and XhoI
Variant 4	CP040 without Spacer 2	98	KpnI and XhoI
Fragment 1	Spacer 1 only	12	KpnI and XhoI
Fragment 2	CSA UTR only	44	KpnI and XhoI
Fragment 3	Spacer 2 only	28	KpnI and XhoI
Fragment 4	CSB UTR only	23	KpnI and XhoI
Fragment 5	Spacer 3 only	19	KpnI and XhoI
CB	Chicken β-actin promoter	583	KpnI and XhoI

To determine whether all 5 fragments of the mini-promoter were needed for optimal performance, the capabilities of the full-length CP040 mini-promoter (Variant 1, all five fragments) (Figure 1A) were compared to those of three variant sequences (Variants 2–4), each containing 4 selected fragments (Figure 1B), individual fragments, and the well-characterized, strong and ubiquitous (but much longer) CBA promoter. A dual luciferase assay was performed in two different cell types: human embryonic kidney cells (HEK293s) and human induced pluripotent stem cell-derived neural progenitor cells (NPCs). Each of the 4 variants was cloned into a promoter-less vector (pGL4.20), upstream of a firefly luciferase (Fluc) reporter gene. The renilla luciferase (Rluc) gene provided in a separate plasmid pRL-CMV was used to normalize transfection efficiency by co-transfection (Figure S2). In HEK293 cells, the fold change in promoter activity from Variant 1 (full-length CP040) was significantly higher compared to the empty vector control. Variant 2 (missing Spacer 1) showed significantly lower activity compared to Variant 1, and Variants 3 and 4 showed negligible activity (Figure 1C). In NPCs, the fold change in Variant 1 (full-length CP040) activity was significantly higher compared to the empty vector and Variant 1. The other fragments and variants showed no activity (Figure 1D). These results demonstrate that the full-length CP040 mini-promoter (Variant 1) is the optimal design, with reduced performance when key fragments are omitted.

Once we established the excellent in vitro performance of full-length CP040, we then evaluated full-length CP040 activity in vivo. The CP040 mini-promoter was cloned upstream of the eGFP reporter gene in an AAV inverted terminal repeat (ITR) containing vector and packaged into the AAVDJ capsid. A dose of 1 × 10¹³ vg/kg of AAVDJ-CP040-eGFP was administered to WT neonatal mice via systemic, i.v. temporal vein injections (Figure 2A). Six weeks later, the mice were euthanized and tissues collected to assess AAV vgs, eGFP transcript levels, and eGFP protein expression. The highest transcript levels were observed in the cortex and cerebellum followed by the spinal cord, liver, and heart (Figures 2B and 2C). There were very few vgs or transcripts observed in the spleen, quadriceps (quad), and kidneys. Immunohistochemistry was performed to co-label eGFP expression with neuronal (NeuN), astrocyte (GFAP), and oligodendrocyte (olig2) markers (Figure 2D), and double-positive cells for each were quantified (Figure 2E). 12.75% ± 7.70% cells were GFP+/NeuN+, 12.10% ± 6.55% were GFP+/GFAP+, and 15.33% ± 6.26% were GFP+/olig2+. These data demonstrate that CP040 can drive transgene expression in vivo across various CNS cell types.

Figure 2.

In vivo characterization of the full-length CP040 promoter packaged in the AAVDJ capsid

(A) 1 × 10¹³ vg/kg of AAVDJ-CP040-eGFP or vehicle (1x PBS) was injected into the temporal vein of wild-type FVB/NJ neonatal mice (postnatal day 1–2). Necropsies were performed on the mice at six weeks of age, and tissue samples were analyzed for vg content and eGFP transcript levels. (B) Graph presenting vg copy numbers in different organs—cortex, cerebellum, kidneys, spleen, quadriceps, spinal cord, liver and heart. (C) Graph showing the fold change in eGFP transcript levels demonstrating promoter strength in the same organs as in (B). Data represent mean ± SEM, N = 10 mice, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, ns, not significant by one-way ANOVA. (D) Representative IF images showing eGFP expression along with neuronal (NeuN), astrocyte (GFAP), and oligodendrocyte (olig2) markers. Scale bar = 100 uM (E) Cells found to be double-positive for eGFP and one of the other markers (NeuN, GFAP, and olig2) were quantified from the IF images captured. Data presented as mean ± SEM, N = 3 mice, ∗∗∗∗p < 0.0001 by one-way ANOVA.

Once successful in vivo promoter activity for CP040 was confirmed, the efficiency was further evaluated by comparing CP040 with larger but well-established, strong ubiquitous promoters like CBA and EF1α, as well as a neuronal-specific promoter human synapsin 1 (hSYN) promoter (Figure 3A). Each of the four promoters driving eGFP expression was packaged in AAVDJ capsids, and a dose of 1 × 10¹² vg/kg per mouse (a lower dose due to volume constraints from these direct administration routes and their limited injection space) was administered into 6-week-old adult WT FVB/NJ mice via ICV and IT injections (Figure 3B).

Figure 3.

Schematic of the overall experiment design for promoter comparison study

(A) Viral vectors used in the study. (B) Six-week-old wild-type FVB/NJ mice (C) were injected with AAVDJ-CBA-eGFP, AAVDJ-EF1α-eGFP, AAVDJ-hSYN-eGFP, AAVDJ-CP040-eGFP, or vehicle (1x PBS) through the ICV or IT route. (D and E) Mice were euthanized 9 weeks after AAV injection, and tissue analysis done for viral vg and eGFP transcript and protein analysis for promoter comparison.

Nine weeks later, the mice were euthanized, with selected organs harvested for quantification of vg and promoter activity levels via transgene expression analyses. As our goal was to primarily target CNS tissues, specific brain regions were dissected and processed along with liver and kidney. As expected, since all four promoters were packaged in the same AAV serotype (AAVDJ) using the same doses, the vgs for each were effectively the same (Figures 4A–4D, 4F, and 4G) or had subtle differences (Figure 4E) across all tissues analyzed for each route of administration (ICV and IT).

Figure 4.

AAV vg quantification in mouse tissue samples

gDNA was isolated, and vg numbers were calculated per nanogram of gDNA for (A–E) neurological tissues (hippocampus, hypothalamus, cortex, cerebellum, and spinal cord), (F) liver, and (G) kidney. Data are presented as mean ± SEM of the values from six mice; ∗p < 0.05, ∗∗p < 0.01; ns, not significant (based on one-way ANOVA).

To establish the efficiency of the CP040 mini-promoter, eGFP transgene expression in specific brain regions and organs we hope to de-target was determined and compared across all cohorts. Total RNA was isolated from each tissue, and eGFP expression levels were assessed by quantitative PCR (Figure 5). Following IT AAV administration, the fold change in eGFP transcripts using the CP040 mini-promoter was comparable to all three established promoters in the hypothalamus, hippocampus, and spinal cord (Figures 5A, 5B, and 5D). In the cortex and cerebellum, CP040-mediated eGFP expression levels were significantly higher than seen with the CBA, EF1α, and hSYN promoters (Figures 5C and 5E). In the ICV administration cohorts, eGFP transcript levels from the CP040 mini-promoter were lower than those obtained from either of the ubiquitous CBA and EF1α promoters and equivalent to the hSYN promoter in the hippocampus and hypothalamus (Figures 5A and 5B). In the cortex, CP040 eGFP expression was equivalent to those obtained with the EF1α and hSYN promoters and significantly lower than those obtained with the CBA promoter. In cerebellum, CP040 eGFP expression was equivalent to that obtained with the CBA promoter and significantly lower than that obtained with the EF1α and hSYN promoters (Figure 5E). The transcription profiles of eGFP were also evaluated in the de-targeted liver and kidneys for both delivery routes. In these organs, CP040 mini-promoter-mediated expression was found to be similar to that of the tissue specific hSYN and significantly less than observed for both the ubiquitous CBA and EF1α promoters (Figures 5F and 5G). These data suggest that with regard to transcription levels, CP040 is comparable in targeted tissues to these well-established promoters despite its much smaller size while performing similarly (low expression) to the tissue-restricted hSYN promoter in off-target tissues.

Figure 5.

eGFP transgene transcript levels in tissues

Total RNA was isolated and reverse transcribed to cDNA. Fold changes in eGFP cDNA expression levels were calculated compared to vehicle controls for (A–E) neurological tissues (hippocampus, hypothalamus, cortex, cerebellum, spinal cord), (F) liver, and (G) kidney. Data are presented as mean ± SEM of the values from six mice; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant (one-way ANOVA).

Mouse brain IF staining was performed to quantify and more precisely determine the protein expression ability of these promoters within specific neurological cell types. Co-immunostaining to detect eGFP transgene expression and molecular markers for neurons (NeuN), astrocytes (GFAP), and oligodendrocytes (olig2) was performed on fixed brain slices from three mice representing each cohort. To quantify the strength of each promoter, the numbers of eGFP/NeuN, eGFP/olig2, and eGFP/GFAP double-positive cells among the total number of neurons, oligodendrocytes, or astrocytes (respectively) were determined following imaging of coronal mouse brain sections from mice treated using either the ICV route or the IT administration route. The images obtained for analyses were determined following random image sampling of coronal mouse brain sections (Figure S3A).

Following ICV administration, all four promoters yielded similar general distributions of expression in the three cell types across brain regions (Figure 6 and S4–S7). eGFP expression was detected in 9.52% ± 5.25% (CBA), 10.00% ± 5.06% (EF1α), 14.52% ± 4.72% (hSYN), and 13.98% ± 4.41% (CP040) of NeuN+ cells (neurons) (Figure 6C). eGFP expression was detected in 14.42% ± 5.52% (CBA), 8.34% ± 5.29% (EF1α), 7.50% ± 3.84% (hSYN), and 10.58% ± 5.72% (CP040) of GFAP⁺ cells (astrocytes) (Figure 6F). eGFP expression was detected in 10.17% ± 4.79% (CBA), 9.96% ± 5.43% (EF1α), 16.23% ± 5.04% (hSYN), and 12.78% ± 5.30% (CP040) of olig2+ cells (oligodendrocytes) (Figure 6I).

Figure 6.

Relative transduction of different cell types between promoter cohorts in the brain following ICV delivery

Representative IF images (eGFP, green; cell type-specific marker, red) from selected brain regions as indicated in the brain overview image from each promoter cohort are shown. eGFP protein expression was compared to a neuronal cell marker (NeuN) (B and Figure S4A), astrocyte marker (GFAP) (E and Figure S4B), and oligodendrocyte marker (Olig2) (H and Figure S4C). The co-localization of each cell type marker and eGFP was quantified across cohorts through blinded analysis. For each mouse tissue, five non-overlapping images were acquired from across each of five tissue sections and evaluated by a trained, blinded observer. The percentages of eGFP/cell-specific marker double-positive cells are presented in the right-hand panels for (C) neurons, (F) astrocytes, and (I) oligodendrocytes. Data are presented as mean ± SEM of the values from three mice; ∗p < 0.05, ∗∗p < 0.01; ns, not significant (based on one-way ANOVA). Scale bars, 100 μm.

Through the IT delivery route, all four promoters yielded similar general distributions of expression in the three cell types across brain regions (Figure 7 and S8–S11). eGFP expression was detected in 9.36% ± 4.16% (CBA), 10.14% ± 4.55% (EF1α), 13.54% ± 4.95% (hSYN), and 10.79% ± 5.18% (CP040) of NeuN+ cells (Figure 7C). eGFP expression was detected in 14.01% ± 4.44% (CBA), 10.62% ± 3.73% (EF1α), 11.13% ± 5.25% (hSYN) and 11.40% ± 4.21% (CP040) of GFAP⁺ cells (Figure 7F). eGFP expression was detected in 10.98% ± 5.29% (CBA), 8.65% ± 4.97% (EF1α), 12.72% ± 3.19% (hSYN), and 12.18% ± 3.46% (CP040) of olig2+ cells (Figure 7I). An overall summary of CP040 promoter strength compared to the CBA promoter (based on the IF results) shows that CP040 performed equivalently to the hSYN promoter and slightly better than the EF1α and CBA promoters in these cell types (Figure 8).

Figure 7.

Relative transduction of different cell types between promoter cohorts following IT delivery

Representative IF images (eGFP, green; cell type-specific marker, red) from the brain region indicated in the brain overview image from each promoter cohort are shown. eGFP protein expression was compared to a neuronal cell marker (NeuN) (B and Figure S8A), astrocyte marker (GFAP) (E and Figure S8B), and oligodendrocyte marker (Olig2) (H and Figure S8C). The co-localization of each cell type marker and eGFP was quantified across cohorts through blinded analysis. For each mouse tissue, five non-overlapping images were acquired from across each of five tissue sections and evaluated by a trained, blinded observer. The percentages of eGFP/cell-specific marker double-positive cells are presented in the right-hand panels for (C) neurons, (F) astrocytes, and (I) oligodendrocytes. Data are presented as mean ± SEM of the values from three mice; ∗p < 0.05, ∗∗p < 0.01; ns, not significant (based on one-way ANOVA). Scale bars, 100 μm.

Figure 8.

Heatmap showing the strength of the promoters compared to CBA

The average numbers of eGFP/cell-specific marker double-positive cells were determined, and fold changes were calculated with respect to CBA for the two routes of AAV delivery (ICV and IT).

Based on these characterization and comparison studies of our CP040 mini-promoter with the well-established but much longer CBA, EF1α, and hSYN promoters, it is evident that our 126-bp CP040 mini-promoter successfully drives transgene expression in neurons, astrocytes, and oligodendrocytes following IT or ICV administrations at levels that are comparable to the longer promoters.

Discussion

A wide variety of constitutive, tissue-restrictive, and inducible promoters have been used to drive transgene expression in gene therapy vectors. Regardless of the promoter type, its size is an important consideration when delivering large transgenes, bi-cistronic proteins, split inteins, or complex cargo using AAV.^40,41,42,43 Several strong constitutive promoters have been developed and used in gene therapy studies over the years. The ubiquitous 800-bp cytomegalovirus (CMV) early enhancer promoter⁴⁴ has been used in several gene therapy clinical trials.^45,46 A major limitation of this promoter is that depending upon the cell type and experimental conditions, it can be prone to silencing over time.^47,48,49,50 The CBA promoter containing the CMV early enhancer sequence and a truncated SV40 late 16S intron used for comparison in our study is another widely used constitutive promoter. CBA can drive strong gene expression in many cell types and is incorporated into the FDA-approved SMA gene therapy design.¹⁷ A limitation to this promoter is its size (∼1,600 bp), which limits its use to only those transgenes that are small enough to fit within the remaining vector space (∼3,400 bp). Another consideration regarding these promoters is the viral origin of the CMV sequences. Recent FDA guidance based on long-term follow-up of human subjects in gene therapy trials has recommended the use of non-viral promoters to reduce the potential for genotoxicity.⁵¹ Thus, current gene therapy vector designs being developed for translation into future human clinical trials are typically focused on the use of synthetic or mammalian promoter sequences.

The neuron-specific promoter derived from human synapsin 1 (hSYN, 448 bp) has been used for targeted transgene expression in mature neurons in various animal models with the AAV9 and AAV-PHP.B capsids to improve gene delivery to the CNS and was used in this study as the neuronal-specific promoter for comparison to our CP040 mini-promoter.⁵² While hSYN is a good choice for CNS transduction, it is also too long for many canonical transgene cDNAs. The RPE65⁵³ and the VMD2⁵⁴ promoters are widely used in retinal gene therapies and are specific to the retinal cells.⁵⁵

Other constitutive mammalian promoters include human β-actin (hACTB. 1,600 bp) and human elongation factor-1α (hEF-1α, 1,172 bp), which have each demonstrated sustained transgene expression for extended periods of time in vivo.^{32,48,56,57,58,59} While these are both well-characterized in the literature, they are also large. Small enhancer-less promoters such as polyubiquitin C (UbC, 403 bp) and methyl CpG-binding protein-2 promoter (MeCP2, 229 bp) could accommodate the size of the larger genes but may provide expression levels that are inadequate for gene therapies.^44,58,60 Finally, the 200-bp synthetic JeT promoter (constructed as a chimeric promoter from fragments of the SV40, human β-actin, and ubiquitin C promoters⁶¹) or the UsP promoter (identical to JeT but includes an expression-boosting intron sequence, 326 bp) have been used in gene therapies being developed for Tay-Sachs disease and neuronal ceroid lipofuscinosis type 7.^62,63

In this study, we designed and characterized a novel mini-promoter, CP040, for AAV-mediated gene expression in the CNS. Given the critical need for compact, yet effective, promoters in gene therapy, our work provides an important contribution to the field by expanding the available promoter options for CNS-directed gene delivery.^64,65 We designed the mini-promoter sequence using 5′ UTR sequences from human ERCC6 and ERCC8 promoter sequences due to the needs of our CS gene therapy programs. Based on information provided by the Human Protein Atlas (HPA), ERCC8 and ERCC6 exhibit low tissue specificity with similar RNA transcript levels across tissues throughout the body based on consensus (combined HPA and Genotype Tissue Expression - GTEx) RNA sequencing datasets. Our examination of endogenous ERCC6, ERCC8, and GAPDH showed similar transcript levels for each across all the tissues that were assessed (no tissue specificity) with ERCC6 and ERCC8 levels being roughly equivalent and GAPDH being higher in each tissue (Table S1).

Using in silico analysis, the sequence was trimmed to 126 bp and found to include many transcription-factor-binding sites that are expected to enhance promoter function (Figure S1) using the online PROMO tool. The Transcription factor Affinity Prediction web tool was also used to scan the ERCC8 and ERCC6 promoter fragments. The ERCC8 fragment includes a binding profile for the aryl hydrocarbon receptor nuclear translocator (ARNT) complexed with hypoxia-inducible factor 1-alpha (HIF1α) ARNT-HIF1α heterodimer. This transcription factor is in the basic-helix-loop-helix (bHLH) family and is a master regulator of the cellular response to hypoxia. In conditions of low O₂ levels, HIF1α dimerizes with ARNT and the complex binds to hypoxia response elements in DNA sequences to activate genes involved in angiogenesis, glucose metabolism, and cell survival.⁶⁶ The ERCC6 fragment was found to include several transcription factor-binding sites (GATA2, ETS1, YY1, NFIC, ELF5, and ELK4) that are known to be enriched in many promoter and enhancer sequences and are broadly involved in a variety of fundamental biological processes.

Our characterization studies determined that the full-length version of our promoter is the most effective variant and highlighted the importance of specific transcription factor-binding sites in achieving robust transgene expression. In vivo experiments demonstrated that AAVDJ-CP040-eGFP administration (1 × 10¹³ vg/kg body weight) via neonatal i.v. delivery resulted in strong expression in CNS tissues, cortex, cerebellum, and spinal cord and underscored CP040’s potential as a broadly applicable mini-promoter therapies in which the CNS is a primary target.

The AAV doses selected for these studies were needed to ensure consistent transduction across the CNS and various cell types based on our previous experience with AAVDJ.⁸ Also, neonatal mice undergo rapid cell division as they grow, and over time this dilutes the total number of AAV episomes within cells. Ongoing and future studies in disease models will include multiple AAV doses and time points for i.v. administrations to more fully assess promoter behavior.

Compared to other well-characterized, widely used promoters, we found that CP040 provided comparable strength using two distinct AAV administration routes (IT and ICV). The performance of CP040 is particularly relevant for applications requiring efficient expression of long transgenes, which often require compact regulatory elements due to the packaging constraints within AAV vectors.³⁶ Our findings suggest that CP040 has the potential to enhance the design of gene therapies targeting neurodegenerative diseases and other CNS disorders where sustained and efficient expression of larger transgenes or other sequences is required. Additionally, due to the incorporation of ERCC8 and ERCC6 promoter elements, which drive the expression of genes that are active throughout all cells, there is the potential for broad applicability of this promoter beyond neurological tissues.

While CP040 exhibited robust performance in human cells and mouse models in this study, future studies in human iPSCs differentiated into a variety of cell types, large animal models, and human clinical trials will further validate its efficacy and potential. Also, our study primarily focused on eGFP as a reporter gene. Additional studies using specific transgenes in various disease models will be necessary to assess the promoter’s efficacy in specific therapeutic contexts.⁶⁷ Further work integrating CP040 with advanced gene editing tools such as CRISPR-Cas9 and various base editor technologies may also provide novel opportunities for precision medicine approaches in neurological disorders.⁶⁸

Our longest time point was 9 weeks post-administration, which is shorter than the typical human gene therapy timeline where the goal is to provide therapeutic expression for many years if not the lifetime of the patient. Thus, studies incorporating CP040 into therapeutic vectors for long-term expression evaluations will be needed to validate the long-term durability of this promoter. Finally, potential immune responses to prolonged CP040-driven transgene expression remain to be investigated.^69,70

In conclusion, we designed and characterized a 126-bp CP040 mini-promoter that has the potential to be the prototype for a new generation of mini-promoters for AAV gene therapies. The major advantage of this promoter is its compact size. An additional benefit is that it consists entirely of human-derived sequences, with no viral promoter or enhancer elements. This promoter offers a balance between broad expression potential (owing to the inclusion of ERCC6 and ERCC8 promoter regions that normally support DNA repair in all cell types, and whose disruption leads to multisystem disease with prominent neurological involvement) and enhanced activity in the CNS due to the presence of transcription factor-binding sites that are highly active in (although not exclusive to) neuronal tissues.

Materials and methods

Cell culture

All cell lines were maintained at 37°C in a 5% CO₂ incubator. Human embryonic kidney (HEK293) cells were grown in high-glucose DMEM +10% FBS (fetal bovine serum). NPCs were maintained in NPC expansion media (47 mL DMEM/F12 [#11320033, Gibco], 1 mL B27 supplement [#17504044, Gibco], 0.5 mL N2 supplement [#17502048, Gibco], 0.5 mL NEAA [#11140050, Gibco], 0.5 mL GlutaMax [#35050061, Gibco], 0.5 mL antibiotic antimycotic [#15240062, Invitrogen], 30 ng/mL EFG [#GMP100-15, Peprotech], and 30 ng/mL FGF [#GMP100-18B, Peprotech]) on basement membrane (Matrigel, #354277, Corning)-coated tissue culture ware.

Dual luciferase assay

Dual luciferase assays were performed using the dual luciferase reporter assay system (#E1910, Promega). All reagents were prepared as described by the manufacturer. From a confluent master plate of healthy iPSC-derived NPCs or HEK293 cells, 2 × 10⁶ cells were seeded in 6-well plates. The following day, the media was changed to serum-free media, 3 h prior to transfection. Using Lipofectamine 3000 (L3000001, Invitrogen), the cells were transfected with 500 ng of either pGL4.20 empty plasmid, pGL4.20 CP040 variants or fragments, or pGL4.20 CBA plasmid along with pRL-CMV control plasmid (pGL4.20: pRL should be 50:1).⁷¹ The media was replaced with normal growth media after 6 h of transfection. After 24 h, the cells were harvested in 500 μL per well of 1x passive lysis buffer and lysed for 20 min on an orbital shaker. The lysate was collected in 2-mL tubes and centrifuged briefly. In a white-bottom 96-well plate, 20 μL supernatant was added with 50 μL luciferase assay substrate (LAS), and luminescence was measured with a 10 s equilibration time and 10 s integration time. This was followed by the addition of 50 μL of Renilla luciferase reagent and firefly quenching (1x Stop & Glo), a 10 s equilibration time, and measurement of luminescence with a 10 s integration time. The respective values were subtracted from non-transfected values.⁷² For each sample, the ratio of firefly (first luminescence reading) to Renilla (second luminescence reading) luciferase activity (Fluc/Rluc) was calculated and presented as fold change compared to the empty vector (pGL4.20) transfection.

Construction of plasmids

A plasmid harboring the CBA promoter driving the eGFP was constructed and used for self-complementary AAV-CBA-eGFP packaging in a previous study.⁷³ This plasmid (pscCB-eGFP) was used as a template to generate new plasmids encoding different promoters located upstream of the eGFP sequence. Briefly, the human EF1α promoter, human SYNapsin 1 gene promoter (hSYN), and CP040 mini promoter sequences were amplified from their source plasmids by using corresponding primer pairs and were replaced with the CBA promoter in template plasmid (pscCB-eGFP). For this purpose, (1) EF1α promoter was amplified using forward primer (FP): 5′-CGCACGCGTAAGCTTTGCAAA-3′ and reverse primer (RP): 5′-CAGGGATCCCCAATACCAGTGTGCAGATCTTGGC-3′, (2) hSYN promoter was amplified using FP: 5′-ATACGCGTAGTGCAAGTGGGTTTTAGGACC-3′ and RP: 5′-AAGGATCCCTGCGCTCTCAGGCACGACA-3′, and (3) CP040 mini promoter was amplified using FP: 5′-ATACGCGTCTCGCCCTGCTCAAAGGCGT-3′ and RP: 5′- AAGGATCCGGAACAGCAGAGGTCAGTCGGT-3’. All amplified promoter sequences and the pscCB-eGFP plasmid were digested with MluI and BamHI at 37°C overnight. Both digested PCR products and the pscCB-eGFP plasmid were run on agarose gels and appropriate bands were isolated (Zymoclean Gel DNA Recovery Kit, #SMP11-300, Genesee Scientific). Purified PCR products were separately cloned between MluI and BamHI restriction enzyme sites in the template plasmid by using Rapid DNA Ligation Kit (#REFK1422, Thermo Scientific). Ligation products were transformed into E. coli SURE competent cells (#200238, Agilent), separately. The accuracy of the cloning was confirmed by Sanger sequencing of the plasmids. Plasmids were amplified in bacteria and isolated using the NucleoBond Xtra Maxi kit (#740414.50, Macherey-Nagel).

Production and purification of AAVDJ vectors

pscCB-eGFP, pscEF1α-eGFP, pschSYN-eGFP, and pscCP040-eGFP constructs were each packaged into AAVDJ vectors using the previously described triple transfection method.^74,75 Briefly, the plasmids encoding (1) the ITR-flanked gene of interest, (2) adenovirus helper genes, and (3) Rep/Cap sequences were mixed with PEI (#23966-1, Polysciences). Adherent HEK293 cells were then co-transfected with PEI:Plasmid in equimolar ratio. 72 h after transfection, cells were harvested, subjected to a freeze-thaw cycle, and treated with Benzonase (#E8263, Sigma-Aldrich) at 37°C for 1 h. After a centrifuge-based clarification step, the supernatant was collected and purified by iodixanol (#D1556, MilliporeSigma) gradient ultracentrifugation, followed by ion-exchange chromatography using HiTrap Q HP (Cytiva) anion exchange chromatography column with Bis-Tris propane-MgSO₄ buffer. Subsequently, the vector preparation was concentrated into 10 mM Tris-100 mM sodium citrate buffer (pH 7.0) using centrifugal spin concentrators (#AP2015010, Orbital Biosciences).

AAV titer determination

For determination of viral genome titers of vector preparations, DNase-resistant particle titers were performed as previously described.^74,76 10 μL aliquots from purified AAV samples were incubated with TURBO DNase (#ref. 4022G, Thermo Fisher Scientific) for removal of pDNA contaminants. DNase-treated mixtures were then treated with Proteinase K (#AM2546, Invitrogen) to release the ITR-flanked recombinant genome packaged within the AAVDJ vector. Finally, the released viral genome was purified using DNA Clean & Concentrator 25 (#11-305C, Zymo). Eluted viral genomes were quantified by qPCR with a TBGreen Advantage (#S4748, Takara Bio). For the determination of vg titer of the scCBA-eGFP construct, the following primers specific to CBA promoter were used: FP: 5′-TCCCATAGTAACGCCAATAGG-3′ and RP: 5′-CTTGGCATATGATACACTTGATG-3’. For the determination of vg titer of the scEF1α-eGFP and schSYN-eGFP constructs, the bovine growth hormone polyadenylation (BGH-PolyA) signal-specific primers BGH-FP 5′-TAGTTGCCAGCCATCTGTTG-3′ and BGH-RP 5′-GCGATGCAATTTCCTCATTT-3′ were used. For the determination of vg titer of the scCP40-eGFP construct, the following primers specific for the CP040 promoter were used: FP: 5′- GTCCCTCCAAACCACGTCTT-3′ and RP: 5′- GGAACAGCAGAGGTCAGTC-3′.

Animals

All husbandry and animal-related procedures were performed under a protocol approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Minnesota (UMN). Five-week-old WT FVB/NJ (stock# 001800) mice were ordered from Jackson Laboratory. They were acclimatized to the vivarium for 1 week prior to initiation of the study. The mice were maintained in the UMN Research Animal Facility at 25°C and 50% humidity in a 14-h light and 10-h dark cycle and fed normal chow and water ad libitum.

AAV delivery

Neonatal temporal vein injections, i.v.

Neonatal mice were injected through the superficial temporal vein at 1–2 days of age. The mice were anesthetized by placing them on ice for less than 5 minutes to induce hypothermia. Once the neonates were anesthetized, a 29.5G short needle syringe was used to deliver AAV solution or vehicle (1x PBS) in a total volume of 20 μL directly into the left temporal vein.⁷⁷ The mice were then warmed on a heating pad to assure recovery prior to being returned to the cage with the dam.

IT injections

IT injections were performed at 6 weeks of age on five male and five female WT mice.⁸ The mice were injected with either vehicle (1x PBS) or with each promoter: AAVDJ-CBA-eGFP, AAVDJ-EF1α-eGFP, AAVDJ-hSYN-eGFP, and AAVDJ-CP040-eGFP. For IT injections, hair was removed from a small area on the lower back using depilatory cream and the injection site was scrubbed. Mice were anesthetized with isoflurane induction at 3% and maintained between 1% and 2% during the duration of the injection. 10 μL of AAV (∼1 × 10¹² vg per kg body weight per mouse) directly via lumbar puncture was injected with a Hamilton syringe (26G).^8,78,79,80

ICV injections

Bilateral ICV injections were performed as previously described.⁸ Briefly, each mouse was anesthetized and the hair was removed between the ears. The head was scrubbed and fixed into a stereotaxic apparatus. Through a small incision, the lateral ventricles, −0.5 mm in the anteroposterior axis, ±1 mm in the mediolateral axis, and −2.3 mm in the dorsoventral axis (coordinated according to Allen Brain Atlas for a ∼2-month-old mouse), were located. Once the needle tip (Hamilton 26G) was placed inside each ventricle, 5 μL of vehicle, AAVDJ-CBA-eGFP, AAVDJ-EF1α-eGFP, AAVDJ-hSYN-eGFP, or AAVDJ-CP040-eGFP was injected slowly into each ventricle with a flow rate of 1 μL per minute.⁸¹ The liquid was allowed to diffuse for 2 min before removing the needle from each ventricle. Groups of five male and five female mice were injected at 6 weeks of age for each promoter group. The final AAV dose was ∼1 × 10¹² vg/kg body weight per mouse. The incision was closed with tissue adhesive (3M Vetbond, 3M Animal care products, USA). Finally, meloxicam (OstiLox, VETone, UK) at a dose of 0.5 mg/kg was injected subcutaneously as a post-operative analgesic, and once daily for 72 h afterward. Animals were monitored for 5 days post-surgery according to IACUC guidelines at the UMN.

Vector genome quantification

AAV vg copy numbers from tissues collected 9 weeks post-AAV administration were quantified by qPCR after extraction of total gDNA using a DNeasy Blood and Tissue Kit (Qiagen). Tissues were mechanically homogenized using ceramic beads in a bead basher (Benchmark BeadBlaster24) with provided buffer and Proteinase K. gDNA isolation was performed according to manufacturer instructions. Vg content in each tissue, normalized to 100 ng of gDNA, was determined via qPCR method using a Taqman probe specific for the eGFP reporter sequence (Assay ID: Mr00660654, Thermo). A standard curve was created using eGFP plasmid DNA dilutions with known copy numbers, to determine absolute copy numbers in the sample gDNA, as described previously.⁸

Reporter gene transcript analysis

Total RNA was isolated from mouse tissues 9 weeks post-AAV administration using the Quick-RNA MiniPrep kit (R1055, Zymo Research) as described previously.⁸ Tissues were mechanically homogenized using ceramic beads in a bead basher (Benchmark BeadBlaster24) with provided RNA lysis buffer. 300 ng of total gDNA-free-RNA was used for cDNA synthesis using the High-Capacity RNA-to-cDNA Kit (#4388950, Applied Biosystems). This cDNA product was used to set up qPCR with eGFP probes (Assay ID: Mr0066065, Thermo) and Taqman Fast Advanced Master Mix (Thermo) in the QuantStudio 3 (Applied Biosystems) thermal cycler. Ribosomal RNA 18s amplification (Assay ID: Mm03928990_g1, Thermo) was used as an internal control on the same cDNA. The fold change in eGFP transcripts with respect to 18s was calculated using the ΔΔC_T method.

Immunohistochemistry

Animals prepared for immunohistochemistry studies were fixed using intracardial perfusion as previously described.⁸ Mice were deeply anesthetized with isoflurane and perfused with ice-cold sterile 1x PBS (pH 7.4) followed by ice-cold 4% paraformaldehyde (PFA) in 1x PBS. Post-fixation, target tissues were removed and placed in 4% PFA overnight for 12 h. The following day, the tissues were dehydrated and embedded into paraffin blocks. 7 -μm-thick sections were cut on pre-charged slides (Superfrost Plus slides #12-550-15, Fisher Scientific). For IF, tissue sections were rehydrated as previously described⁸ and then rinsed in water. The slides were kept in 1x PBS for 5 min, then permeabilized for 30 min at room temperature in 1x PBS containing 0.3% Triton X-100, and then incubated for 1 h in blocking solution (1x PBS containing 3% BSA, 2% normal donkey serum, and 0.15% Triton X-100). Primary antibody dilutions in the blocking solution were prepared and incubated on samples overnight at 4°C. The following primary antibodies were used: mouse monoclonal anti-GFP, 1:50 (Invitrogen, #A11121, specificity confirmed by lack of labeling in the absence of AAVDJ-eGFP treatment), rabbit anti-OLIG2, staining of oligodendrocytes, 1:100 (Proteintech, #13999-1-AP), rabbit anti-NeuN, staining of mature neurons, 1:100 (Proteintech, #26975-I-AP), and rabbit anti-GFAP, 1:100 (Novus Biologicals, #NB300-141, exclusive staining of astrocytes). After a 5-min 1x PBS rinse, sections were incubated for 1 h at room temperature with combinations of Alexa Fluor (1:500) (AF568 goat anti-rabbit, #A11036 and AF488 goat anti-mouse, #A11029)-conjugated secondary antisera (Life Technologies, Invitrogen). After final rinses, sections were cover-slipped using DAPI-containing mounting media (ProLong Diamond Antifade Mountant, Invitrogen #P36962). Tissues from 1x PBS-injected WT mice were also processed with the AAVDJ-CBA-eGFP-, AAVDJ-EF1α-eGFP-, AAVDJ-hSYN-eGFP-, and AAVDJ-CP040-eGFP-treated tissues and were included in all immunohistochemical experiments as controls. These PBS control images (Figure S3) were used to identify and avoid areas of mouse brain autofluorescence due to known lipofuscin accumulation⁸² in region of interest selections.

Microscopy and image analysis

Leica DM5500 B epi-florescent microscope and LAS-X software were used to collect images from IF tissue slides. Images were collected using the same microscope settings for comparing tissues within ICV- and IT-injected mice cohorts. Similarly, for all promoters and injection site groups in target-treated tissues, image adjustments including contrast, brightness, and color were performed in parallel. Five non-overlapping images were acquired across 3 sections for each mouse tissue, which were evaluated by a trained and blinded observer. Brain cell types were outlined based on visualization with cell-type-specific antibodies, and only cells with visible nuclei were counted. The total number of GFP-positive cells using the cell counter plugin within the ImageJ software were counted. The percentage of GFP-positive cells was calculated from the total number of each cell type.

Statistical analysis

For vg and RNA transcript analyses, tissues from n = 6 mice from each group were used. For all the IF studies, tissues from n = 3 mice from each group were analyzed. The data are presented as mean or percentage ±standard error of the mean (SEM) from at least three animals unless otherwise noted. No data were excluded from analyses. Blinded analyses were performed on coded IF assay images by three study investigators A.L.D, A.H., and O.W. One-way ANOVA comparisons were followed by Bonferroni post hoc test or comparisons were made using the Student’s t test unless indicated otherwise. The figure legends indicate numbers of animals used in each experiment (n), independent experiments, or biological replicates. p values were calculated by the indicated statistical tests using GraphPad Prism (v.10.4.1) software.

Data and code availability

The novel promoter sequence described and characterized in this study along with all other data supporting the findings of this study, including plasmid maps and image analysis scripts, are available from the corresponding author upon reasonable request and will be shared in accordance with institutional guidelines and material transfer agreements.

Acknowledgments

This work was supported by the US Food and Drug Administration Center for Biologics Evaluation and Research, contract/grant number: R01FD007483 (C.A.P. and G.A., MPI), NIH NINDS 1R61 NS135019-01A1 (C.A.P.), as well as generous support from the Viljem Julijan Association for Children with Rare Diseases (Slovenia). A portion of the mini-promoter CP040 characterization work (M.C.) was supported by the funding from the Office of Discovery and Translation, Clinical and Translational Science Institute (CTSI) at UMN. The authors thank the Center for Immunology and the Institute of Translational Neuroscience Imaging Core and Dr. Jason Mitchell for the microscope training, all at the University of Minnesota. The authors appreciate the assistance of Venessa Agyapong and Frances Sehneah with RNA and gDNA isolation. The graphical abstract and Figures 2A, 3, 6A, 6D, 6G, 7A, 7D, 7G, S2A, S3A, and S3B were partially created using BioRender, under a license permitting their use in publications (Created in BioRender. Pacak, C. (2025) https://BioRender.com/4lc09l4).

Author contributions

M.C. and C.A.P. designed the experiments, analyzed the data, and wrote the manuscript with input and editing from P.B.K., G.A., A.L.D., and J.A.T. Recombinant viral vector production, purification, and its quality controls were performed by O.F.D. and G.A. AAV administrations were performed by M.C., A.L.D., and F.(E.)K., M.C. and A.L.D. performed quantitative PCRs for vg and transcript analysis. M.C. and A.L.D. performed histology preparations, and M.C. performed IF assays. A.L.D., A.H., and O.W. quantified the IF assays blinded. C.A.P. and G.A. acquired funding. All authors reviewed a draft of the manuscript and provided input.

Declaration of interests

C.A.P. and M.C. are inventors on a patent application (PCT/US2024/037893) for CP040.