Inherited retinal dystrophies (IRDs) constitute a spectrum of conditions characterized by both clinical and genetic heterogeneity, affecting approximately 1 in 3,000 individuals (Hanany et al., 2020). This clinical diversity ranges from non-progressive diseases like night blindness to progressive conditions such as retinitis pigmentosa (RP) (García Bohórquez et al., 2021; Perea-Romero et al., 2021). The identification of over 300 genes conclusively linked to IRDs (RetNet, https://web.sph.uth.edu/RetNet/; accessed on 13 January 2024) also exhibits the complex nature of these conditions, marked by extensive genetic heterogeneity and variable expressivity. Notably, different variants within the same gene can lead to a wide range of clinical presentations (Chen et al., 2018; Bianco et al., 2023), and conversely, similar clinical manifestations may result from variants in different genes (García Bohórquez et al., 2021; Perea-Romero et al., 2021; Smirnov et al., 2021). Additionally, evidence supports the significant role of gene modifiers in human eye diseases (Meyer and Anderson, 2017; Li et al., 2021), introducing an added layer of complexity to the diagnosis and understanding of IRDs.

The TULP1 gene serves as an example of the heterogeneity observed in IRDs. The link between TULP1 and autosomal recessive IRDs was first established in 1998 (Banerjee et al., 1998; Hagstrom et al., 1998). Since then, TULP1 biallelic variants have been extensively associated with various forms of IRDs, including non-syndromic RP, Leber congenital amaurosis, cone dystrophy, and rod–cone dystrophy (Ullah et al., 2016; Bodenbender et al., 2023). To date, the LOVD database (www.lovd.nl/gene, accessed on 13 January 2024) records a total of 117 unique TULP1 variants. Among these, 46 are classified as variants of uncertain significance (VUS), while 106 are categorized as likely pathogenic or pathogenic.

TULP1 encodes Tubby-like protein 1, which is a member of the TULP protein family (North et al., 1997; Nishina et al., 1998). TULP proteins have been well-documented for their critical roles in the development and function of the central nervous system (Ikeda et al., 2002). Despite displaying different expression patterns, all TULP proteins share a conserved C-terminal region of approximately 200 amino acids known as the tubby domain. TULP1, in particular, is exclusively expressed in the retina, primarily within the cytoplasm of retinal photoreceptor cells (Ikeda et al., 1999). It is assumed that TULP1 is involved in various protein–protein interactions and the intracellular protein transport, including rhodopsin and vesicle transport to and from the outer segments of photoreceptor cells (Hagstrom et al., 2001).

Bodenbender et al. (2023) reported the largest TULP1-related IRD cohort to date. This research established a connection between biallelic TULP1 variants and a wide range of clinical phenotypes, emphasizing the remarkable heterogeneity in TULP1-related retinal dystrophy. The authors suggested that protein misfolding could be a key factor contributing to the variable clinical presentations resulting from TULP1 gene variants. However, the intricate mechanisms underlying how TULP1 misfolding leads to specific clinical phenotypes still remain unknown (Boggon et al., 1999; Ikeda et al., 2002).

In our current research, we explore the clinical and molecular characteristics of a patient with biallelic TULP1 variants presented with a unique pattern of macular degeneration and periarteriolar vascular pigmentation. Our primary focus lies on the identification of both a newly identified and a previously reported variant within the TULP1 gene. This study aims to broaden the understanding of the relationship between TULP1 and IRDs.

TULP1 is a crucial retinal protein for intracellular protein transport within photoreceptor cells. It has been associated with a wide range of retinal disorders including RP and early-onset RP, Leber congenital amaurosis, cone dystrophy, and rod–cone dystrophy (Banerjee et al., 1998; Hagstrom et al., 1998; Lobo et al., 2016; Woodard et al., 2021; den Hollander et al., 2007; Jacobson et al., 2014). Despite the diversity of associated conditions, consistent differences in clinical presentation attributed to specific TULP1 variants, or their impacts on the protein have not been reported (North et al., 1997; Lobo et al., 2016; Remez et al., 2020; Jia et al., 2022). Notably, it has been suggested that patients carrying homozygous or compound heterozygous TULP1 pathogenic variants, particularly those affecting the tubby domain and/or resulting in a loss-of-function (LOF) effect, typically exhibit a severe, early-onset form of retinal dystrophy (den Hollander et al., 2007; Jacobson et al., 2014; Jia et al., 2022; Mataftsi et al., 2007).

Here, we describe a patient presenting a novel TULP1 gene variant (c.822G>T), which has been predicted to disrupt the splicing process. The variant is located within the disordered domain, for which structural information is currently lacking (Boggon et al., 1999; Lobo et al., 2016). In a hypothetical scenario where the protein is synthesized, it is possible that the tubby domain might be absent. Our minigene splice assays demonstrated that the c.822G>T variant induces exon 8 skipping, resulting in the formation of a premature stop codon p.(Gly240Glufs*109). Alternatively, there is a possibility that, instead of exon skipping, the last eight base pairs of exon 8 are omitted, thereby leading to a premature stop codon p.(Gly272Glufs*109). In both scenarios, the introduction of a premature stop codon is likely to activate the NMD mechanism, ultimately resulting in the absence of protein production from this allele (Figure 4B). Similarly, Bodenbender et al. (2023) performed minigene assays for the c.1495+1G>A and c.1496–6C>A TULP1 variants. In both cases, the variants were predicted to induce frameshift variants and premature termination codons, leading to the probable degradation of the mutant transcript via the NMD mechanism. The first variant achieved this by inducing exon 14 skipping and the insertion of 17 amino acids p.(Ala442Profs*18), while the second variant activated a cryptic acceptor site p.(Pro499Leufs*143). Further functional studies would be required in retinal cells to establish the pathogenic mechanism, considering that splicing is specific to the tissue.

On the other hand, the c.1376T>C, p.(Ile459Thr) variant induces a substitution at amino acid position 459, replacing an isoleucine with a threonine. This change entails a shift from a hydrophobic side chain to a polar one, which could potentially affect protein stability or induce misfolding (Lobo et al., 2016). The variant p.(Ile459Thr) is located in the tubby domain of TULP1, which has previously been shown to exhibit DNA-binding activity and function as a transcription factor (Carroll et al., 2004). Missense variants situated in the tubby domain are expected to accumulate at the endoplasmic reticulum, potentially activating the unfolded protein response, which can promote apoptosis and photoreceptor cell death (Jia et al., 2022). Interestingly, the variant c.1376T>C has been previously reported in trans with the c.1112+2T>C variant in a patient with non-syndromic RP (Wang et al., 2014). Although both our patient and the patient reported by Wang et al. (2014) present the same missense variant (c.1376T>C) affecting the tubby domain and another variant that causes LOF, their phenotypes are markedly distinct, illustrating the significant variability in patients with TULP1 variants.

Our case exhibits bull’s eye maculopathy characterized by perivascular pigmentation and increased autofluorescence around retinal vessels (Figure 2). This phenotype corresponds to observations detailed by Al-Hindi et al. (2022), who reported a similar distinctive phenotype in two patients with cone dysfunction and a perivascular pattern of retinal degeneration. Case 1 in Al-Hindi et al. (2022) displays a clinical and FAF appearance remarkably similar to our patient, both in progression and test outcomes. However, notable differences exist, such as a greater degree of macular atrophy in our patient, potentially linked to the older age of our subject (51 years vs. 19 years). Remarkably, case 1 is a homozygous carrier of the p. (Gly363Arg) missense variant impacting the tubby domain (Figure 5). The second case in Al-Hindi et al. (2022) shares similarities with our case in terms of the bull’s eye maculopathy. The FAF also suggests periarteriolar hyperautofluorescence. Interestingly, case 2 shares a genotype composition similar to our patient, carrying a missense variant (p.(Cys523Tyr)) affecting the tubby domain, alongside a LOF variant (p.(Lys274ArgfsTer36)). In contrast to cases described by Al-Hindi et al. (2022), our patient lacks pigment clumps surrounding vessels and shows no signs of perivascular choroidal atrophy. However, the distinct pattern of periarterial degeneration observed in both our case and Al-Hindi et al. (2022) cases has not been documented in other studies.

Bodenbender et al. (2023) reported a total of 17 different TULP1 pathogenic variants, categorized as missense, splice site, and nonsense variants, and one in-frame deletion. All documented variants, except the missense variant p.(Gly266Val), either affected the tubby domain or led to LOF (Figure 5). The p.(Gly266Val) variant, identified in trans with a LOF TULP1 variant, was associated with cone dystrophy in a male patient (P15 in their series) diagnosed at age 40. This suggests that missense variants like p.(Gly266Val), not affecting the tubby domain, may result in a relatively mild reduction of protein activity and a later-onset phenotype. When comparing our case with those in Bodenbender et al. (2023) series, participant P17 stands out as the most analogous, exhibiting bull’s eye maculopathy. Although P17’s retinography only captures the central 50°, there is no apparent perivascular hyperautofluorescence pattern. Similarly, on OCT, mirroring our case, outer segments are visible, although with foveal discontinuity.

In this study, we present a single case of a patient with compound heterozygous TULP1 variants. The prevalence of biallelic pathogenic TULP1 variants is remarkably low, accounting for less than 1% of reported non-syndromic IRD cases, as documented in GeneReviews (Fahim et al., 2023; Kumaran et al., 2023). Within our extensive cohort of over 1,000 pediatric and adult patients diagnosed with IRDs, we have identified an additional case, distinct from the one presented here, with confirmed TULP1 pathogenic variants (internal data). Our findings align with the existing literature, which documents only 78 patients with TULP1-associated IRDs from 44 families (Hagstrom et al., 1998; Ullah et al., 2016; Bodenbender et al., 2023; Woodard et al., 2021; den Hollander et al., 2007). This collective evidence underscores the limited occurrence of such cases, reinforcing the distinctive nature of our reported case.

Additional factors have been suggested to contribute to the variable clinical presentation observed in TULP1-related IRDs, including protein misfolding (Bodenbender et al., 2023), environmental factors or the influence of genetic modifiers (Meyer and Anderson, 2017). MAP1A has been suggested as a modifier gene for TULP1 in a mouse model (Maddox et al., 2012; Youn et al., 2022), offering protective effects and elucidating variations in disease manifestation. However, the analysis of the MAP1A gene in our patient revealed five homozygous variants (Supplementary Table S4) with a high population frequency, predicted to have no impact on the protein. Consequently, these variants may not significantly influence the patient’s phenotype. Nevertheless, exploring the MAP1A gene on all patients with biallelic variants in TULP1 would be of great interest to determine if there are specific variants associated with a protective effect or those linked to a more severe phenotype.

To gain a deeper insight into the pathogenesis and variations in the phenotype, further investigations should be conducted using animal models of the retina. Additionally, these studies can contribute to the development of gene therapies, similar to those that have been successfully developed for other IRDs (Russell et al., 2017). Efforts have been made for TULP1, although a supplementation therapy trial targeting photoreceptors in Tulp1^−/− mouse retinas demonstrated limited effectiveness (Palfi et al., 2020). This study is still crucial to increase our understanding of the phenotypic traits of the disease. This knowledge forms the basis for early diagnosis and future therapies.

In conclusion, we described a patient with biallelic TULP1 variants, displaying an atypical perivascular pattern of retinal degeneration. Furthermore, we illustrated that the variant c.822G>T induces aberrant splicing, resulting in either exon 8 skipping or the loss of the last 8 bp of exon 8, both culminating in LOF. The molecular characterization of the c.822G>T variant allowed its reclassification, establishing its pathogenic effect. This highlights the importance of conducting functional studies on variants situated outside canonical splice sites that still show potential to impact splicing.