Tissue-Engineered Endothelial Keratoplasty with Controlled Cell Density: Toward Super TEEKs

Abstract

Over the past 20 years, endothelial keratoplasty procedures have revolutionized the treatment of corneal endothelial disorders. These conditions have now become the leading indication for corneal transplantation in Western countries and account for half of all donor cornea usage. Despite their undeniable success, the global shortage of donor tissues and major disparities between nations justify the development of alternatives to donor grafts. Cell therapy using injections of suspended endothelial cells has proven effective, and tissue-engineered endothelial keratoplasty (TEEK), comprising a membrane coated with cultured endothelial cells, is under development to better mimic the native endothelial graft. Our team utilizes a femtosecond-laser-cut lens capsule disc as a bioengineering scaffold, taking advantage of this novel tissue’s biocompatibility, transparency, curvature, and availability. In the present study, we provide proof of concept, in 12 TEEKs, that it is possible to control the final endothelial cell density (ECD) by varying the seeding density per mm². Cell characterization was performed through morphometric analysis of the endothelial mosaic stained with anti-NCAM (a lateral membrane marker used as a differentiation marker), using the CellPose artificial intelligence algorithm specifically trained for in vitro endothelium segmentation. Five criteria related to pleomorphism, polymorphism, and elongation were combined into a single endothelial quality score. The median cell viability at 28 days of culture, assessed by Hoechst 33342 and Calcein-AM staining, reached 98% (range: 83–99%). The median viable ECD (number of live cells per surface unit) in the highest-density group was 3.245 cells/mm² (range: 2.778–3.753), paving the way for the bioengineering of supra-physiological TEEKs, or “super TEEKs”.

Impact Statement

The manufacturing process of tissue-engineered endothelial keratoplasty (TEEK) enables control over endothelial cell density (ECD), allowing for the production of “super TEEKs’: grafts with supra-physiological ECDs that are better equipped to withstand surgical stress and may offer greater survival in recipients.

Keywords

corneal bioengineering corneal endothelium human corneal endothelial cells (hCECs)human anterior lens capsule viable tissue-engineered endothelial keratoplasty (TEEK)

Introduction

Diseases of the corneal endothelium are the leading indication for corneal transplantation in Western countries. Their standard curative treatment is endothelial keratoplasty (EK), which consists of removing the central 8 or 9 mm of the pathological endothelium and replacing it with a disc of the same diameter dissected or cut from a donor cornea. Clinical results from numerous case series have long confirmed the extremely favorable risk-benefit ratio of EK: anatomical restoration allowing for the potential of full visual recovery (20/20), long graft survival with a minimally invasive procedure, very low per- and postoperative complication rates, and also a very low rejection rate. Its main drawback remains, to this day, the complete dependence on corneal donation, which is globally insufficient and extremely unequal from one country to another.¹

With the introduction in Japan of the very first approved corneal endothelial cell therapy,² the treatment of corneal endothelial diseases has entered a new era: that of Advanced Therapy Medicinal Products (ATMPs). The first product developed is a suspension of human corneal endothelial cells (hCEC) cultured in vitro from a small number of donor corneas under the age of 30. The goal of these innovative treatments is to replace donor corneas in these indications.

In addition to the injection of suspended cells, bioengineered corneal endothelial grafts could represent a promising therapeutic alternative. Tissue-engineered endothelial keratoplasty graft material (TEEK) precisely replicates the endothelial graft from a donor: it is composed of a transparent membrane, equivalent to Descemet’s membrane (a basal membrane equivalent for hCEC), covered with a single layer of hCEC to form a functional neo-endothelium.

The human anterior lens capsule has already been used as a culture substrate for corneal endothelial cells (CECs) of human and porcine origin^3–6 An original femtosecond laser cutting process allows us to reproducibly obtain an 8-millimeter diameter lens capsule disc (LCD), regardless of the lens diameter, which varies in humans from 9 to 11 mm, mainly depending on age. Its efficacy has been demonstrated in rabbits, where a xeno-TEEK (LCD and hCEC) restored corneal transparency in a model of complete endothelial injury.⁵

The relatively simple manufacturing process allows for control of the number of hCEC seeded on the LCD, then for measurement of the endothelial cell density (ECD in cells/mm²) and assessment of cell quality at the end of culture, prior to delivery. Several questions remain unresolved: does this manufacturing process allow for the production of TEEK with a predetermined ECD, and in particular, TEEK with a very high, supra-physiological ECD, which could allow for prolonged graft survival? In this study, we seeded LCDs with three increasing ECD levels and compared the number of viable hCEC as well as cell morphology after 28 days of culture.

Method

Human corneas and crystalline lens collection

Crystalline lenses were collected during corneal retrieval (PFS16-010) (Supplementary Table S1). Special care was taken to preserve the lens capsule by avoiding direct pinching. The lenses donors’ age was 73 ± 9 (56; 90) (mean ± standard deviation [min; max]). Lenses were stored in 20 mL of organ culture medium (CorneaMax, Eurobio, Les Ulis, France) at 31°C for up to 62 days (20 ± 13 [1; 62]) to evaluate the feasibility of long-term storage before processing.

Primary hCEC cultures were established using corneas retrieved for research, as authorized by the French Biomedicine Agency (PFS16-010 and PFS21-001). Corneas were stored in organ culture medium (CorneaMax) at 31°C prior to hCEC isolation. Corneas from three donors aged 15, 83, and 88 years were used. The death-to-harvest time was less than 24 h. All samples were handled in accordance with the tenets of the Declaration of Helsinki.

Preparation of the anterior lens capsule disc

Lenses were placed in a sterile watertight Petri dish, a custom holder machined in polyetheretherketone to allow accurate centration during laser cutting. The anterior capsule was cut using a femtosecond laser prototype system. The source was a Satsuma HP (Amplitude, Bordeaux, France) emitting at 1030 nm, with a 328 fs pulse duration. Laser pulses passed through a Galvanometer scanner (Scanlab, Puchheim, Germany) and were focused with an f-theta lens. The focus was a Gaussian-shaped 8 μm diameter spot (1/e²), spaced 5 μm apart. Pulse energy was 4 μJ. The capsule was cut to produce an anterior LCD of 8 mm in diameter with 3 peripheral asymmetrical marks (2 aligned, 1 offset). This layout, originally used in EK, ensured proper tissue orientation. A circular x–y scan was applied at 250 kHz, with 5 μm spacing along the circle edge, as shown in Figure 2. This was repeated vertically (z-axis) in 2 μm steps over 4.6 mm centered on the capsule. The high-density tubular pattern ensured full cutting of the capsule.

LCDs were then dissected using microsurgical tools under a binocular microscope (SZ61, Olympus, Tokyo, Japan). After staining with 0.06% trypan blue (VisionBlue, DORC, Zuidland, The Netherlands) to facilitate visualization, the LCD was detached from the lens using toothless forceps. LCDs were spread on a glass slide with sterile water to verify their integrity under a microscope.

They were placed in a glass vial (11553542, Fisher Scientific, Loughborough, UK) containing 4 mL of sterile water (B230531, Versylen, Fresenius, Bad Hamburg, Germany) and 1/100 antibiotic-antimycotic (15240–062, Gibco, Saint Louis, MO, USA). These tubes were then agitated at room temperature (RT) for 3 days to ensure the decellularization of the lens epithelial cells. The LCDs were then stored at 4°C in sterile water until use.

Cell culture

All cell cultures were performed using the “peel and digest” method previously described^7,8 and adapted for the first clinical applications of endothelial cell therapy by injection.² The composition of the three cultures media was detailed (Table 1).

Table 1.

Composition of the Three Media Used in the Corneal Endothelial Cell Culture Process. The Passage Medium Was the Expansion Medium Supplemented with Y-27632

Use	Component designation	Company	Reference	Final concentration
Digestion	Optimem-I	Gibco	11058-021
	CaCl₂	Sigma	C5670	200 µg/mL
	Ascorbic acid	Sigma	A5960	20 µg/mL
	Collagenase A	Gibco	A12177	2 mg/mL
Expansion	Optimem-1	Gibco	11058-021
	CaCl₂	Sigma	C5670	200 µg/mL
	Ascorbic acid	Sigma	A5960	20 µg/mL
	Fetal calf serum	Eurobio	CVFSVF00-01	8%
	Gentamycine	Gibco	15710,049	50 µL/mL
	Chondroïtine Sulfate 2%	Wako	0341462	0.8 mg/mL
	Epidermal Growth factor h (EGFh)	Gibco	PHG0311	5 ng/mL
	SB203580. Adezmapimod. p38 MAP kinase inhibitor	Selleckchem	S107612	1.1513 µg/mL (10 µM)
	SB 431542. TGF-beta inhibitor	Selleckchem	S1067	0.1151 µg/mL (1 µM)
Passage	Expansion medium
Passage	+Y-27632. Rock Inhibitor	Selleckchem	S1049	10µM

Each cornea was treated independently. Briefly, the Descemet’s membrane (DM) was peeled off using forceps, under an operating microscope (Universal S3, ZEISS, Germany), using a “no-touch” technique to obtain the largest possible surface of endothelium without trabeculum. The DM was then exposed to 0.06% Trypan blue (Visionblue, DORC, Zuidland, The Netherlands) for 30 s to reassess endothelial quality (absence of large areas of dead cells). It was then quickly rinsed in CorneaMax and incubated in a well of a 24-well plate (3524, Costar, Corning Incorporated, USA) containing 500 µL of digestion medium for 16 h in a humidified incubator at 37°C and 5% CO2. 2.5 mL of passage medium was then added to stop the enzymatic reaction of collagenase A, and the contents of the well were transferred to a conical tube (64199, Caplugs Evergreen, Dutsher). After centrifugation (D37520, Sigma, Germany) at 200 g for 5 min, the supernatant was removed, and the cell pellet was resuspended in 100 µL of passage medium. The dissociation of the cells was completed by performing about 10 successive pipetting steps. The 100 µL were then seeded into a well of a 12-well plate with 380 mm² surface area (3513, Costar, Corning Incorporated, USA) in 1.5 mL of passage medium. The passage medium was replaced after 24 h with expansion medium. The total volume of each well was then fully renewed once a week with expansion medium.

To perform the passages, the cells were rinsed with Dulbecco’s phosphate-buffered saline (PBS) without calcium or magnesium (LM-S241, Biosera, France), incubated at 37°C for 15–30 min in a 5X TriplE solution (A12177-01, TriplE 10X, Gibco, Life Technologies Limited, UK), and then mechanically detached by performing five to six successive pipetting steps. The suspension was centrifuged in a conical tube at 200 g for 5 min with 2.5 mL of passage medium. The cell pellet was resuspended in 100 µL of passage medium by pipetting several times to dissociate the cells. Passages were all performed at a surface ratio of 1:2 for 24-well to 12-well plates and 12-well to 6-well plates, and at a ratio of 1:2.5 for 6-well plates, in order to double the culture surface at each passage (Fig. 1).

FIG. 1.

Schematic of the passaging strategy. P: passage, Pn: a subsequent undefined or hypothetical passage (e.g., P4, P5, etc.), where n > 3. The four plate wells images were adapted from Image provided by Servier Medical Art (http://smart.servier.com/), licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

All wells, except for the initial digestion step, were coated with a laminin 511 solution (892012, iMatrix Nippi, Incorporated, Tokyo, Japan) diluted in PBS (without calcium or magnesium) at 0.25 µg/cm² and left in place as recommended by the supplier for 1 h at 37°C. The entire coating solution was then replaced with passage medium (volume varying depending on the chosen well plate) and placed in the incubator until seeded within the next 20–30 min. No drying time was performed during the coating step, in accordance with the supplier’s instructions.

Assembly of Tissue-Engineered Endothelial Keratoplasty

The preparation method for the LCDs was the same as recently described.⁶ The LCDs were rinsed 3 times in sterile PBS before being spread, under a microbiological safety cabinet (MSC), on the polyethyleneterephthalate membrane of 113 mm² inserts (665610, Greiner BIO-ONE) that were previously coated with 150 µL of FNC Coating Mix (a solution of fibronectin, collagen, and albumin 0407, Athena ES, Baltimore, MA, USA). To test the influence of the LCD orientation on the growth of hCEC, half of the LCDs were spread in one direction and the other half in the opposite direction (Fig. 2). After 2 h of drying under MSC, the capsules were coated with iMatrix (0.25 µg/cm²) for 1 hour in the incubator at 37°C.

FIG. 2.

Principles of cutting anterior lens capsule discs (LDC), orientation, and seeding of corneal endothelial cells on the anterior or posterior surface.

The hCEC from the three donors (Table 2) with a characteristic endothelial phenotype (joint hexagons without mesenchymal endothelial transition) were selected (Supplementary Fig. S1). Three groups of TEEK were generated by varying the number of seeded cells between 500, 2500, and 4000 cells/mm². The hCEC were harvested in the same manner as during a standard passage. The cell concentration per µL was determined (in duplicate) using an automatic counter (TC20, BioRAD, Hercules, CA, USA), and the volume was adjusted to obtain the desired cell density. For each group, the hCEC were seeded on both the anterior and posterior sides of the LCD to analyze the importance of this polarity. The TEEK were then incubated at 37°C and 5% CO₂ for 28 days, following the same expansion medium renewal schedule as for primary hCECs: the first renewal 24 h after seeding, then once a week with 1 mL in the bottom of the wells and 600 µL in the inserts.

Table 2.

Characteristics of the Corneas Used for Primary Cultures of Endothelial Cells. ECD before Use, Corresponded to the ECD of the Primary Culture before Cell Harvesting and Seeding on the Lens Capsule Discs. Culture Assignment Corresponded to the Number of TEEKs Performed Using Each Culture

Sex/age (years)	Death-to-retrieval time (h)	Initial corneal ECD (cells/mm²)	In vitro ECD before use (cells/mm²)	Passage	Culture assignement
Sex/age (years)	Death-to-retrieval time (h)	Initial corneal ECD (cells/mm²)	In vitro ECD before use (cells/mm²)	Passage	TEEK 500	TEEK 2500	TEEK 4000
M/15	18	2809	1865	P9	2	2	2
F/83	16	2114	1500	P4	2	2	2
F/88	22	2337	1600	P4	1	1	1

M, male; F, female; ECD, endothelial cell density.

Final quality controls

After 28 days of storage, two parameters were measured: 1/the viable ECD (number of viable hCECs per unit area, or viable corneal endothelial cell density [vECD]) and 2/the morphometry of cells labeled with an anti-NCAM used as a reference marker for well-differentiated CECs without EndoMT.^9,10

The measurement of vECD was obtained after double staining with Hoechst 33342/Ethidium homodimer, using a protocol adapted from Pipparelli et al.¹¹ Briefly, the TEEKs adherent to the membrane of the culture inserts were rinsed with 500 µL of PBS then incubated for 45 min at RT, protected from light, with a solution of Hoechst at 5 µg/mL (B2261, Sigma, Saint Quentin Fallavier, France) and Calcein-AM at 4 µM (FP-FI9820, Interchim, Montluçon, France) diluted in Opti-MEM I. The TEEKs were then observed directly in the inserts under a macroscope (macrozoom microscope, MVX10, OLYMPUS, Tokyo, Japan) at ×1 and ×6.3 magnification (Supplementary Fig. S2). The surface covered by viable cells (Calcein-positive) was measured using the Cornea J plugin on FIJI¹² on ×1 macroscope images.

The 12 TEEKs still adherent to the bottom of the inserts (n = 3 TEEK 500; n = 5 TEEK 2500; and n = 4 TEEK 4000; six seeded on the anterior side and six on the posterior side) were then fixed in methanol (A456-212, Fisher Scientific, Loughborough, UK) for 45 min at RT. The immunostaining protocol was similar to the already published protocol.⁹ Briefly, the inserts were rinsed with 500 µL of PBS, then 100 µL of blocking solution, composed of 2% goat serum (11475055, Fisher Scientific) and 2% BSA (A3059, Sigma Aldrich) diluted in PBS, was added into the inserts. The plate with the inserts containing the blocking solution was incubated under gentle agitation at 37°C. After 30 min, the blocking solution was replaced with the primary antibody solution composed of the blocking solution supplemented with hNCAM.I (IgG2b, R&D, MAB24081, R&D Systems, Minneapolis, USA) diluted at 1/500. After 1 h at 37°C under agitation and protected from light, the compartments were gently rinsed three times with 500 µL of PBS. Finally, the secondary antibody goat antimouse IgG2b AlexaFluor 555 (A110, Invitrogen) diluted at 1/1000 in the blocking solution and 4’,6-diamidino-2-phenylindole dihydrochloride (DAPI) at 1 mg/mL (D9542, MilliporeSigma, St. Louis, MO, USA) were incubated for 1 h at 37°C under gentle agitation and protected from light. The insert membrane carrying the TEEK was then carefully cut (Fig. 3.A–C) and mounted flat between a slide and coverslip in a drop of fluorescent mounting medium (Neo-Biotech, MB23.00158.1, Nanterre, France). The TEEKs were observed under an epifluorescence microscope (IX81, Olympus) with CellSens software (Soft Imaging System GmbH, Munster, Germany). Observations were made with the ×4 objective and the multi-image alignment mode to cover the entire surface of the TEEK, then with the ×10 and ×40 objectives in different areas of interest of the TEEK (Supplementary Fig. S3). The number of nuclei per unit area was obtained on Fiji using the macro function deep learning-based nuclei/cell detection and segmentation method (Stardist) from images acquired under the epifluorescence microscope at ×10 with a resolution of 2048 × 2048 pixels (Fig. 3.D). For the calculation of the vECD, the average value from the center and intermediate zone (weighted by the number of nuclei counted) was used: vECD = %Calcein-positive surface × (nuclei/mm² in these two zones).

FIG. 3.

Tissue-engineered endothelial keratoplasty (TEEK) after immunostaining. (A) TEEK adherent to the bottom of an insert. (B) Cutting of the insert membrane without touching the TEEK. (C) Membrane and TEEK ready to be mounted between slide and coverslip. (D) Diagram of the different observation zones on and outside the TEEK.

Cellular segmentation and morphological analysis

As described recently in¹³ we tuned a Cellpose model^14,15 for segmenting NCAM-labeled hCECs. Cellpose is a generalist cell and nuclei segmentation algorithm with multiple pretrained models, each trained on thousands of cells, but which was not adapted to NCAM-labeled hCECs; hence, the need to tune a model specifically for this labeling on these cells, which is largely facilitated by an easy-to-use graphical user interface. We developed a mathematical analysis Python script (v3.11.4) to measure five parameters on segmented images: coefficient of variation of cell area (CV in %), adjusted CV (in %), percentage of cells with six neighbors (HEX), HEX-Q (%), and filimorphism (%) to conduct extensive cell polymegathism and pleiomorphism analysis (Fig. 4). We also used an endothelial quality score (EQS) to evaluate the global morphology of TEEK hCECs. This scoring system assigned 50% weight to ECD, as it is the primary criterion used in clinical practice (eye banking), and 50% to morphology, considering that both were equally important. Morphology evaluation was based on the five parameters previously described: HEX, HEX-Q, CV, adjusted CV, and filimorphism, each contributing 1/10 of the EQS. The EQS was derived from Z-scores, which were calculated by the following formula: $Z_{i} = \frac{X_{i} - µ}{σ};$ where $Z_{i}$ was the score value, Xi the raw value, μ the population mean for the given criterion, and σ the standard deviation of the referred population. The higher the EQS, the better the endothelial quality.

FIG. 4.

Example of Cellpose segmentation applied to NCAM labeled hCECs with computed morphometric parameters: CV, Adjusted C.V, HEX, HEX-Q and filimorphism (in %) and Endothelial Quality Score (EQS) for global morphology evaluation.

Statistical analysis

Statistical analysis was performed using IBM SPSS Statistics 30.0.0.0. Given the sample sizes of the 3 groups, comparisons were carried out using the nonparametric Kruskal–Wallis test with Post-Hoc Dunn’s test using a Bonferroni corrected alpha for multiple comparisons (pairwise group comparisons) to evaluate the influence of the LCD side, as well as for the comparison of morphometric criteria, EQS, and vECD.

Experiment

After 28 days of culture, for each of the three TEEK groups, morphological criteria, EQS and vECD did not differ significantly between the hCECs adhering to the LCD and those adhering to the membrane of the culture insert outside the TEEK. Similarly, for each of the three TEEK groups, vECD did not differ significantly between the two sides of the LCDs (data not shown). The data obtained on the anterior and posterior sides were therefore grouped for the subsequent comparisons between the three groups. The data were therefore obtained by analyzing 7739 ± 2528 cells (min 3500–max 11875) per TEEK. All TEEKs showed a continuous endothelium (Fig. 5). As expected, the vECD of the TEEK 500 was significantly lower than that of the other two groups (p = 0.005). None of the five morphometric parameters taken separately differed between the three groups, but the EQS score was significantly lower for the TEEK 500 than for the TEEK 4000 (p = 0.0037). Between the TEEK 2500 and 4000, none of the six criteria differed significantly. All of these results were shown in (Fig. 6).

FIG. 5.

Uniformity and viability after 28 days of tissue-engineered endothelial keratoplasty (TEEK) with three increasing seeding densities. Staining with Hoechst 33342 and Calcein-AM (green fluorescence) directly in their culture insert. Observations under macrozoom microscope (×1). Yellow dotted lines outlining the TEEK. Viability expressed as % of the TEEK surface.

FIG. 6.

Comparison of endothelial morphometry and viable corneal endothelial cell density (vECD) between the 3 groups. Individual values and median (horizontal bar). C.V, coefficient of variation of cell area; HEX, hexagonality; HEX-Q, corrected hexagonality; EQS, endothelial quality score.

Discussion

In the strategy of replacing donor corneas with alternative solutions, TEEK faithfully replicates the descemet membrane endothelial keratoplasty (DMEK) endothelial graft. It consists of a biocompatible support, an LDC in our case,^5,6 on which a monolayer of hCEC adheres, forming a neo-endothelium capable of restoring ad integrum the corneal histology in its entirety. In this pilot study, we demonstrate that it is possible to control the final ECD during the bioengineering process and obtain grafts with supra-physiological ECD.

We used an original method of femtosecond laser capsulotomy that ensures the reproducible production of 8 mm diameter LCDs with minimal manual intervention and a conventional method of hCEC culture.⁶ Note that we have obtained that the lens can now (since November 2024) be harvested for therapeutic purposes from deceased individuals.¹⁶

For the final quality assessment of TEEK, we measured the vECD,¹¹ which stands as the best quantification of the truly useful cellular capital (37 referenced articles since our initial description in 2010^11,17) and used an innovative method based on an AI image analysis algorithm to determine an EQS from a very large number of automatically segmented cells. The EQS can be considered as an automatic and objective assessment of endothelial mosaic quality, usable as a new biomarker, particularly for spotting hCECs in mesenchymal-to-endothelial transition. This new test has currently been trained on immunostaining images of NCAM (destructive test). During the production of clinical batches, it could be used on a sacrificed sample or needs to be adapted to images without staining or with a nontoxic stain to be used on each TEEK. It will complement the critical quality attributes used to release cell batches.¹⁸

This work helps to clarify several critical points of the bioengineering process. When hCECs are seeded at 500 cells/mm², they reach a ECD close to their initial ECD before replanting on the LCDs after 28 days. However, it is possible to achieve a higher final ECD than before replanting by seeding a larger number of cells on the LCDs. This concentration effect allows for modulation of the final quality of the TEEK, but notably, there seems to be a maximum limit, as none of the 4 TEEK 4000 exceeded 3800 viable cells/mm². It is likely that this limit is related to the ECD before replanting (1865 cells/mm² in this series), but this will need to be verified experimentally.

Notably, the achievement of TEEK with a vECD >2000 cells/mm² using donors aged 83–88 years is an important breakthrough, as up until now, clinical-grade cultures have been obtained from donors under 30 years of age².¹⁹

Finally, we have clarified that the orientation of the LCD (anterior or posterior face) does not seem to influence the adhesion and growth of hCECs. To our knowledge, this result has not been known until now.^20–22

TEEK presents several theoretical clinical advantages over the injection of cells in suspension: (1) the presence of a structured basement membrane on which hCECs reform a confluent endothelium; (2) the possibility of maintaining cultures to allow for the maturation of various intercellular junctions. These two characteristics should enable early endothelial function recovery similar to a current DMEK. Furthermore, TEEK transplantation allows for an exact replication of the current surgical procedure, which involves the removal of the patient’s pathological endothelium, whereas it is preserved in the injection therapy; (3) elimination of the risk of a large number of cells being carried into the general circulation (out of the 1 × 10⁶ injected cells, just under half reform an endothelium of 110 mm² with a maximum of 4000 cells/mm²). Even though this situation has not caused any adverse effects in preclinical²³ and clinical^2,18,19 studies of suspension cell injection therapy, this notable difference should facilitate the approval of this ATMP by health authorities; (4) control over the number of adherent cells delivered to the patient without the need to inject a molecule promoting cell adhesion (such as Rho-Kinase inhibitors injected with suspension cells); (5) similarly to DMEK, TEEK should be transportable from the manufacturing site (tissue therapy unit) to the operating room under simple conditions, potentially preloaded and ready to be transplanted.²⁴These various theoretical advantages will, of course, need to be verified by clinical data.

In the clinical practice of endothelial grafts, at the request of surgeons, cornea banks reserve corneas with high ECD for this technique. While the validation threshold for a conventional full-thickness graft was set about 40 years ago at 2000 cells/mm² (a pragmatic threshold balancing satisfactory clinical results and an economically acceptable proportion of donor corneas) and is nearly universal worldwide, it is most often set at 2400 cells/mm² for endothelial grafts, as the preparation and surgical technique involve more manipulations that cause additional cell loss, which varies depending on the difficulties encountered on a case-by-case basis. Here, we show that TEEK bioengineering should allow achieving even higher ECD levels capable of competing with the postoperative ECD of 4000 cells/mm² reported 2 years after treatment with suspension cell injection.¹⁸ This supra-physiological ECD should allow for better resistance to intra- and postoperative events, thereby prolonging graft survival, leading to the concept of the “super TEEK.” Moreover, the ability to customize TEEK production also opens the perspective of personalizing graft allocation to the recipient while optimizing the limited hCEC resource: for example, a TEEK of 1500–2000 cells/mm² could most likely result in excellent visual recovery in an 85-year-old patient or older, as demonstrated by hemi²⁵ and quarter DMEK.²⁶

The main weakness of this pilot study is its limited power. In particular, it was not possible to differentiate between TEEK 2500 and 4000. The accumulation of data on a larger scale, for example, during an industrialization process, will allow for a multivariate analysis of the influence of various parameters concerning cell culture (donor age, ECD before culture, method and duration of preservation, composition of the culture medium, seeding density, and maturation time after seeding).

Conclusion

The bioengineering of endothelial grafts on LCDs allows for the control of the number of viable cells after 4 weeks and, consequently, the creation of a range of TEEKs, potentially enabling personalized allocation to the recipient. It also allows for the production of grafts with supra-physiological ECD. These “super-TEEKs” could have prolonged survival in recipients.

Footnotes

Acknowledgments

The authors would like to thank the Fondation des Aveugles de Guerre (Paris) for their support (1-year PhD scholarship for Ms. Inès Aouimeur). They extend our sincere gratitude to the cornea donors and their families for their generous and invaluable contributions to medical and scientific advancement. They also wish to sincerely thank the Eye Banks of Saint-Étienne, Besançon, and Nantes for their crucial support in providing corneas for research purposes. They also thank the transplant coordination teams of the CHU of Saint-Étienne and Nantes, as well as the laboratory of anatomy at the Faculty of Medicine of Saint-Étienne.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Authors’ Contribution

Conception: I.A., Z.H.E.; A.M., P.G., G.T., C.M.; Experimentation: I.A., L.P., S.F., G.B., T.S., G.T., S.Xxx; Interpretation of Results: I.A., Z.H.E., C.M., G.T., C.M.; Drafting: I.A., Z.H.E., P.G., G.T., C.M.; Editing: A.M., P.G., G.T., C.M.; Final approval: all authors.

Supplemental Material

References

Gain

, Jullienne

, He

, et al. Global survey of corneal transplantation and eye banking. JAMA Ophthalmol, 2016; 134(2):167–173; doi: 10.1001/jamaophthalmol.2015.4776

Kinoshita

, Koizumi

, Ueno

, et al. Injection of cultured cells with a ROCK Inhibitor for Bullous Keratopathy. N Engl J Med, 2018; 378(11):995–1003; doi: 10.1056/NEJMoa1712770

Spinozzi

, Miron

, Bruinsma

, et al. Evaluation of the suitability of biocompatible carriers as artificial transplants using cultured porcine corneal endothelial cells. Curr Eye Res, 2019; 44(3):243–249; doi: 10.1080/02713683.2018.1536215

Kopsachilis

, Tsinopoulos

, Tourtas

, et al. Descemet’s membrane substrate from human donor lens anterior capsule. Clin Exp Ophthalmol, 2012; 40(2):187–194; doi: 10.1111/j.1442-9071.2011.02678.x

Crouzet

, He

, Ben Moussa

, et al. Tissue engineered endothelial keratoplasty in rabbit: Tips and tricks. Acta Ophthalmol, 2022; 100(6):690–699; doi: 10.1111/aos.15081

Ben Moussa

, Parveau

, Aouimeur

, et al. Femtosecond laser cutting of human crystalline lens capsule and decellularization for corneal endothelial bioengineering. Bioengineering (Basel), 2024; 11(3):255.

Okumura

, Kakutani

, Numata

, et al. Laminin-511 and -521 enable efficient in vitro expansion of human corneal endothelial cells. Invest Ophthalmol Vis Sci, 2015; 56(5):2933–2942; doi: 10.1167/iovs.14-15163

, Okumura

, Sato

, et al. Corneal endothelial cell therapy: Feasibility of cell culture from corneas stored in organ culture. Cell Tissue Bank, 2021; 22(4):551–562; doi: 10.1007/s10561-021-09918-8

Bartakova

, Alvarez-Delfin

, Weisman

, et al. Novel identity and functional markers for human corneal endothelial cells. Invest Ophthalmol Vis Sci, 2016; 57(6):2749–2762; doi: 10.1167/iovs.15-18826

10.

Frausto

, Swamy

, Peh

GSL

, et al. Phenotypic and functional characterization of corneal endothelial cells during in vitro expansion. Sci Rep, 2020; 10(1):7402; doi: 10.1038/s41598-020-64311-x

11.

Pipparelli

, Thuret

, Toubeau

, et al. Pan-corneal endothelial viability assessment: Application to endothelial grafts predissected by eye banks. Invest Ophthalmol Vis Sci, 2011; 52(8):6018–6025; doi: 10.1167/iovs.10-6641

12.

Bernard

, Campolmi

, He

, et al. CorneaJ: An ImageJ plugin for semi-automated measurement of corneal endothelial cell viability. Cornea, 2014; 33(6):604–609; doi: 10.1097/ico.0000000000000114

13.

Travers

, Coulomb

, Aouimeur

, et al. Investigating the role of molecular coating in human corneal endothelial cell primary culture using artificial intelligence-driv. bioRxiv, 2025; doi: 10.1101/2025.04.15.648927

14.

Stringer

, Wang

, Michaelos

, et al. Cellpose: A generalist algorithm for cellular segmentation. Nat Methods, 2021; 18(1):100–106; doi: 10.1038/s41592-020-01018-x

15.

Pachitariu

, Stringer

. Cellpose 2.0: How to train your own model. Nat Methods, 2022; 19(12):1634–1641; doi: 10.1038/s41592-022-01663-4

16.

Arrêté du 21 novembre 2024 fixant la liste mentionnée à l’article R. 1241-2-1 du code de la santé publique des tissus et des cellules pour lesquels le prélèvement sur une personne décédée présentant un arrêt cardiaque et respiratoire persistant est autorisé. 2024. Available from: https://www.legifrance.gouv.fr/jorf/id/JORFTEXT000050668079

17.

, Pipparelli

, Manissolle

, et al. Ex vivo gene electrotransfer to the endothelium of organ cultured human corneas. Ophthalmic Res, 2010; 43(1):43–55; doi: 10.1159/000246577

18.

Ueno

, Imai

, Tomioka

, et al. Comprehensive combined analysis of physician-initiated phase II and III clinical trials on a cultured human corneal endothelial cell product for treating bullous keratopathy. Jpn J Ophthalmol, 2024; 68(6):660–668; doi: 10.1007/s10384-024-01123-w

19.

Numa

, Imai

, Ueno

, et al. Five-Year follow-up of first 11 patients undergoing injection of cultured corneal endothelial cells for corneal endothelial failure. Ophthalmology, 2021; 128(4):504–514; doi: 10.1016/j.ophtha.2020.09.002

20.

Spinozzi

, Miron

, Lie

, et al. In Vitro evaluation and transplantation of human corneal endothelial cells cultured on biocompatible carriers. Cell Transplant, 2020; 29; doi: 10.1177/0963689720923577

21.

Yoeruek

, Saygili

, Spitzer

, et al. Human anterior lens capsule as carrier matrix for cultivated human corneal endothelial cells. Cornea, 2009; 28(4):416–420; doi: 10.1097/ICO.0b013e31818c2c36

22.

Van den Bogerd

, S

, Zakaria

. Characterizing human decellularized crystalline lens capsules as a scaffold for corneal endothelial tissue engineering. J Tissue Eng Regen Med, 2018; 12(4):e2020–e2028; doi: 10.1002/term.2633

23.

Koizumi

, Sakamoto

, Okumura

, et al. Cultivated corneal endothelial transplantation in a primate: Possible future clinical application in corneal endothelial regenerative medicine. Cornea, 2008; 27 (Suppl 1):S48–S55; doi: 10.1097/ICO.0b013e31817f2298

24.

Wojcik

, Parekh

, Romano

, et al. Preloaded DMEK with endothelium outward: A multicenter clinical study using DMEK rapid device. Cornea, 2024; 43(1):38–44.

25.

Birbal

, Hsien

, Zygoura

, et al. Outcomes of Hemi-Descemet membrane endothelial keratoplasty for fuchs endothelial corneal dystrophy. Cornea, 2018; 37(7):854–858; doi: 10.1097/ico.0000000000001578

26.

Birbal

, Ni Dhubhghaill

, Baydoun

, et al. Quarter-Descemet membrane endothelial keratoplasty: One- to two-year clinical outcomes. Cornea, 2020; 39(3):277–282; doi: 10.1097/ico.0000000000002127

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

2.66 MB

0.00 MB