Investigative Ophthalmology & Visual Science, Vol. 30, No. 8, August 1989Copyright © Association for Research in Vision and Ophthalmology

Retinal Reattachment of the Primate MaculaPhotoreceptor Recovery after Shorr-Term Detachment

Christopher J. Guerin, Don H. Anderson, Robert N. Fariss, and Steven K. Fisher

The macula of the neural retina from 12 adult rhesus monkeys (Macaca mulatto) was detached fromthe overlying retinal pigment epithelium (RPE) by subretinal injection of a balanced salt solution.Seven days later, the two layers were reapposed by draining fluid from the vitreous cavity andreplacing it with a 3:1 mixture of sulphur hexafluoride gas and air. Animals were sacrificed at 1 hr, 2days and 7 days after detachment, and at periods ranging from 3 to 14 days after reattachment. At 2-7days prior to sacrifice, some eyes received an intravitreal injection of 3H-L-fucose. The eyes were thenfixed for light and electron microscopy (EM), and tissue sections were processed for autoradiography(ARG) or immunocytochemistry. During the 7-day detachment interval, rod outer segments (ROSs)and cone outer segments (COSs) degenerated, but inner segments remained intact and the rest of theretina appeared normal. The apical RPE surface dedifferentiated during the detachment interval. At 3days after reattachment, a regrowth of rudimentary ROSs and COSs had. occurred, but the discstacking was clearly abnormal. ROSs and COSs both showed an increase in length and a tendency toreturn to their normal configurations with increasing time after reattachment. ROSs and COSsregained approximately 40% of their normal lengths after a 2-week reattachment period; however,persistent outer segment abnormalities were frequently found in otherwise well regenerated areas.Autoradiographic results confirmed that new disc membranes were synthesized subsequent to reat-tachment. Newly synthesized rod disc membranes were uniformly labeled using antibodies to bovineopsin. Regenerating outer segments interdigitated with newly formed apical RPE processes, andradiolabeled phagosomes were identified within the RPE cytoplasm by 1 week after reattachment.Proliferation of the RPE cell layer was identified at some locations in all animals, and was stronglycorrelated with a lack of underlying outer segment regeneration. Because of the short detachmentinterval, and the absence of underlying pathology or trauma, the recovery process described hereprobably represents an example of optimum recovery after retinal reattachment. Invest OphthalmolVis Sci 30:1708-1725, 1989

The outer segments of photoreceptor cells are or-ganelles specialized for the absorption and transduc-tion of light quanta into neuronal activity. Degenera-tion of these organelles is one of the most prominentmorphological changes, and one of the most signifi-cant factors in the loss of vision, after retinal detach-ment. Conversely, the capacity of rods and cones toregenerate their outer segments and to reestablishtheir normal anatomical relationship with the retinalpigment epithelium (RPE) is an important factor indenning the limits of visual recovery after retinalreattachment.

From the Neuroscience Research Institute, University of Califor-nia, Santa Barbara, California.

Supported by research grants EY-02082 and EY-00888 to DHAand SKF, respectively, from the National Eye Institute, NationalInstitutes of Health, Bethesda, Maryland.

Submitted for publication: September 19, 1988; accepted March8, 1989.

Reprint requests: Christopher J. Guerin, Neuroscience ResearchInstitute, University of California, Santa Barbara, CA 93106.

Evidence from experimental studies of retinal reat-tachment is generally consistent in demonstratingthat outer segments retain some capacity for regener-ation after detachment.1"4 In addition, the prevailingevidence strongly suggests that detachment durationis an important, and perhaps critical, variable in thatprocess.4 Precise identification of temporal parame-ters, however, has proven to be elusive due to smallsample sizes and the variability inherent in the exper-imental detachment procedure. In studies of humanretinal detachments it has been reported that visualacuity following reattachment can sometimes ap-proach predetachment levels, although the prognosisis much less favorable if the macula is involved.5"9

These results imply that the photoreceptor-RPElayer in the macular region either has a reduced ca-pacity for recovery, or that the high degree of ana-tomical and functional specialization of that arealeads to more severe consequences following injury.

Here we report our initial findings on morphologi-cal recovery of the primate macula after experimental

1708

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Guerin er ol 1709

retinal detachment and reattachment, focusing onthe early stages of recovery following a short detach-ment interval (7 days). The photoreceptor-RPE in-terface undergoes an extensive degree of "remodel-ing" in the first week following reattachment, withslower but continued recovery occurring over thenext 7-day period. There is no indication that themacula is more or less susceptible to the effects ofdetachment than any other retinal region. The pho-toreceptor and RPE abnormalities, previously identi-fied in the detached and reattached cat retina,410 alsooccur in the primate macula. It is highly likely, there-fore, that these abnormalities contribute to the lack ofcomplete visual recovery usually associated withmacular retinal detachments.

Materials and MethodsUnilateral retinal detachments were produced in

12 adult, domestically bred, rhesus monkeys whoseaverage age was > 10 years. Reattachments were per-formed after 7 days. The detachment and reattach-ment intervals, in days, are as follows with the num-ber of animals at each time point given in parenthe-ses: 1:0 (1), 2:0 (1), 7:0 (1), 7:3 (2), 7:5 (1), 7:7 (2),7:14 (3), and 7:30 (1). The boundaries and othercharacteristics of each detachment were recorded ona standard ophthalmic detachment chart at the end ofeach surgery. The eyes were examined periodicallyusing an indirect ophthalmoscope. The animals werekilled by an overdose of barbiturate followed by car-diac perfusion of fixative. All animals were cared foraccording to the ARVO Resolution on the Use ofAnimals in Research.

SurgeryUnilateral retinal detachments were created using a

variation of the technique we published previously.4In brief, animals were anesthetized initially with anintramuscular injection of Ketamine HC1 (Bristol,Syracuse, NY) and Xylazine HC1 (Haver Lockhart,Morris Plains, NJ), and surgical anesthesia wasmaintained with periodic injections of these agents.For local anesthesia a retrobulbar injection of 0.5 cc2% lidocaine HC1 (Vedco, Saint Joseph,. MO) wasgiven. An infusion canula was inserted into a 20-gauge incision made in the inferotemporal pars planaand connected to a bag of lactated Ringers solution(Baxter, Deerfield, IL). A second incision was madesuperior to the first and a complete vitrectomy wasperformed. Using the superior incision a micropi-pette (80-100 /xm tip) was inserted into the vitreouscavity and advanced using a micromanipulator untilit was just above the superior vessel arcade of themacula. Balanced salt solution (Alcon, Fort Worth,TX) was slowly injected through the pipette as it ad-

vanced through the retina. When the pipette reachedthe subretinal space (SRS) a small retinal detachmentformed which could be enlarged to the desired size byregulating the amount of fluid injected. As the vol-ume of injected fluid increased a bullous detachmentformed which separated the neural retina from theRPE. In most of the animals the entire macula wasdetached, and in no case did the detachment includeless than 50% of the macula. Initially the height of thedetachment at its apex was at least several millimetersabove the RPE surface, however, over the course ofthe next 7 days detachment height decreased sponta-neously. At the end of the 7-day interval, detach-ments were not ophthalmoscopically visible. Postop-erative antibiotics, consisting of gentamycin sulfate(Elkins-Sinn, Cherry Hill, NJ), cortiosporin ointment(Burroughs Wellcome, Triangle Park, NC) and peni-cillin (Bristol) were administered and all animals re-covered without complications. Retinae were reat-tached by means of a simple fluid gas exchange with amixture of 75% sulphur hexafluoride gas (Matheson,Cucamonga, CA) and 25% room air. Postoperativeantibiotics were again administered as above. All ani-mals recovered without complications. No attemptwas made to seal the retinal hole and in no case wasredetachment observed.

Light and Electron MicroscopyThe fixation protocol and staining techniques em-

ployed in this study have been previously published.10

Briefly, eyes were fixed by intracardiac perfusion of1% gluteraldehyde and 1% paraformaldehyde inphosphate buffer, and then enucleated. Animals wereentrained to a 12:12 light-dark cycle and euthanized 4hr after light onset. After removal of the anterior seg-ments, eyecups were immersed in the same fixativeovernight. This was followed by post-fixation in os-mium tetroxide, dehydration in graded ethanols andembeddment in Araldite 6005 resin. Sections (1 ^mthick) were cut for light microscopy and stained witha mixture of methylene blue, azure II and toluidineblue in aqueous solution containing Na Borate, andthen counterstained with basic fuchsin and photo-graphed using a Zeiss photomicroscope III. Ultrathinsections for transmission electron microscopy werecut with a diamond knife, stained with uranyl acetateand lead citrate, and then carbon-coated prior toviewing.

Outer segment lengths were measured using a ZeissPhotomicroscope equipped with a XI00 oil immer-sion objective and an ocular micrometer. The proto-col and criteria were essentially the same as thosedescribed previously." Aligned rod and cone outersegments were measured in the maculas of normalcontrol eyes and from the regions adjacent to the

1710 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / August 1989 Vol. 30

Table 1.3H-fucose injection schedule

Detachment period Reattachment period Labeling interval(days) (days) (hr)

7*777*

37

1414

72168

296

* Two animals were injected at these timepoints.

detachment in experimental eyes. Measurementsfrom normal areas of experimental eyes, and normalareas from the opposite control eyes, were pooled andaveraged together. Likewise, measurements takenfrom animals with identical detachment:reattach-ment intervals were also averaged together. Onlyouter segments that were well oriented for their fulllength were measured. At many of the earlier time-points rudimentary outer segments were very disor-ganized or absent entirely; the lengths of very disor-ganized outer segments were not measured. Simi-larly, those photoreceptors with no apparent outersegment were not included in the measurements.

Autoradiography3H-L-fucose (500 /xCi) was injected intravitreally at

the timepoints listed in Table 1. Sections were pro-cessed for electron microscope autoradiography usingthe flat substrate method of Young and Droz12 asoutlined in Anderson, Fisher and Breding.13 Thinsections (60-70 nm) were dipped in Ilford L-4 emul-sion under Na vapor illumination and exposed for4-6 weeks at 4°C. The autoradiograms were devel-oped in complete darkness using either phenidon de-veloper for 1 min at 15°C or Elon-ascorbic acid aftergold latensification.14

ImmunocytochemistryFor the immunocytochemical studies we employed

the indirect, post-embedding technique described inErickson et al.15 Tissue specimens were fixed as de-scribed above, except that immersion fixation was foronly 1 hr, the tissue was stained en bloc using 1%aqueous uranyl acetate, and dehydration was througha series of graded methanols. Tissue specimens wereembedded in LR White resin polymerized at 52°C.Thin sections were placed on nickel grids and prein-cubated in normal goat serum for 15 min beforeovernight incubation in rabbit anti-bovine or rabbitanti-rat opsin diluted 1:400 in phosphate bufferedsaline containing 1% bovine serum albumin (PBS/BSA). After several washes in PBS/BSA, grids wereincubated in goat anti-rabbit IgG-Au (5 nm) (Jann-sen, Piscataway, NJ). After further washing grids werethen stained briefly with osmium tetroxide vapors,

Reynold's lead citrate and uranyl acetate, after whichthey were carbon-coated and viewed in a transmis-sion electron microscope. Control sections were cutfrom nondetached areas of the experimental eyes aswell as from the normal, contralateral eye.

ResultsChanges after Experimental Detachment

Acute (ie, <1 hr) retinal detachments producedsome disruption of the outer segment tips and theapical RPE surface in the detached area; this wasmore severe close to the injection site. Occasional redblood cells were apparent in the subretinal space butthe detached retina surrounding the retinal hole re-mained intact. Figure 1A and B are low-power elec-tron micrographs illustrating the morphological ef-fects of separating the two cell layers using the de-tachment procedure outlined in Materials andMethods. Many outer segments remained intact andattached to their respective inner segments (Fig. IB).Some large packets of lamellar material apparentlydetached from their inner segments as a result of theprocedure and these were situated immediately abovethe intact outer segments. Pigment granules and as-sociated membranous debris derived from the apicalRPE processes were also present in this location. Verylittle outer segment debris was identified in the rest ofthe newly expanded subretinal space. The RPE cellmonolayer was intact although the microvilli on theapical RPE surface were smaller, less numerous andshorter than normal (Fig. 1A). At the edge of thenewly created detachment, there was an abrupt tran-sition zone between detached and attached retina.The detachment height at the transition zone wasquite shallow (ie, <1 retinal thickness) (Fig. 2). Theadjacent attached area appeared normal.

At 2 days post-detachment, the appearance of theretina was similar to the acute detachment. Mostouter segments were still present, but the normal par-allel alignment of the disc membranes at the outersegment bases, the site of new disc assembly, wasdisrupted. The apical RPE surface had a scallopedprofile and was populated by microvillous processesmuch shorter than those that normally interdigitatewith photoreceptor outer segments.

After 7 days of detachment the region of detachedretina that included the macula appeared flat oph-thalmoscopically. Histologically, however, this regionwas still shallowly detached. The height of the de-tachment at 7 days ranged from tens of microns at itsperiphery (Fig. 2) to several hundred microns at itshighest point. At this stage, the photoreceptor outersegments had degenerated almost completely, andthe expanded subretinal space between the photore-

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Guerin er QI i l l

Fig. 1. Electron micro-graph of the RPE (A) andphotoreceptor layers (B) < 1hr after microinjection ofbalanced salt solution intothe subretinal space. (A)The apical microvilli thatinterdigitate with the pho-toreceptor outer segmentswere shorter and less nu-merous after experimentaldetachment. Otherwise, themorphology of the RPE celllayer was normal. (B) Thephotoreceptor layer and therest of the retina also re-mained intact after detach-ment. However, some outersegment fragments ap-peared in the subretinalspace (arrows), while othersremained attached to theirrespective inner segments.Membrane-bound vesiclesand other cellular debris,presumably derived fromdamaged outer segmentsand RPE apical microvilli,were located in the inter-photoreceptor space andimmediately above the in-tact outer segments in theexpanded subretinal space(A, X2400, B, X3625).

ceptors and the RPE was filled with lamellar debris(Fig. 2). Photoreceptor inner segments appeared in-tact, although the organelles usually concentrated ineither the ellipsoid or myoid regions were often co-mingled in both cellular compartments. The numberof mitochondria in the ellipsoid was reduced, individ-ual mitochondria appeared distended, and photore-ceptor nuclei were occasionally displaced from theirusual positions into the inner segment. Phagocyticcells, filled with large amounts of membranous mate-rial, were prominent in the subretinal space near the

distal inner segments. A small amount of disorga-nized membranous material was also evident at thedistal tips of some of the inner segments. The mor-phology of the RPE monolayer was similar to its ap-pearance at 2 days post-detachment.

Outer Segment Recovery after Reattachment

A light micrograph of a parafoveal control areafrom an experimental eye is shown in Figure 3A,illustrating the normal morphological relationship

1712 INVESTIGATIVE OPHTHALMOLOGY b VI5UAL SCIENCE / August 1989 Vol. 30

Fig. 2. The RPE-photoreceptor interface 7 days after experimental retinal detachment seen near the transition zone between attached anddetached retina. Seven days after production of a macular detachment, the retina appeared flat by indirect ophthalmoscopy. However,histologically the retina remained shallowly detached. In this animal, the subretinal space was filled with lamellar debris (inset) almostcertainly derived from degenerating photoreceptor outer segments. Only small disorganized stacks of membrane were apparent at the distaltips of some inner segments (arrow). Phagocytic cells of either RPE or hematopoietic origin were located in the subretinal space (asterisks)(X3300, inset X390O0).

between the photoreceptors and the RPE. The tips ofthe rod outer segments (ROSs) are enveloped by theRPE apical processes and are in close proximity tothe RPE apical cell surface. Extrafoveal cone outersegment (COS) tips are recessed from the RPE apicalsurface and are ensheathed by an organized array ofprocesses, the cone sheath, that terminates near theouter segment base. The cylindrical rod and coneouter segments are closely aligned, as are the myoidand mitochondrial-rich elipsoid regions. Pigmentgranules are positioned near the apical RPE surfaceand sometimes within the apical processes.

Figure 3B through E are a light microscopic surveyof 7-day retinal detachments followed by 3, 5, 7 and14-day reattachment intervals. Figures 4-7 show thecorresponding electron microscopic appearance ofthese retinas at the same four reattachment time-points. Overall, rod and cone outer segments gradu-

ally elongated and tended to regain their cylindricalconfigurations as the reattachment interval length-ened. At 3 days, the earliest reattachment timepointsampled, many photoreceptors appeared to have noouter segment. The outer segments that were presenthad a mean length < 3 jam and their shape was highlyirregular (Figs. 3B, 4, 8A, B). The longest COSs iden-tified were about 3.5 fim (Fig. 8B) and, at this stage,more cones than rods possessed rudimentary outersegments. At 5 days post-reattachment the majorityof inner segments had accompanying outer segments(Figs. 3C, 5). COSs remained quite short (x = 3.2 fim)and showed more irregularities in the organization oftheir outer segment membranes than did rods; meanROS length more than doubled to 6.4 pm. At 7 and14 days mean ROS length increased to 8.7 and 9.9fim, respectively, whereas COS lengths were 6.5 ^mand 7.2 jum at the same timepoints. The increase in

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Guerin er ol 1713

Fig. 3. Light microscopic survey ofthe photoreceptor-RPE interface aftershort-term reattachment. (A) A nor-mal control region from an experi-mental eye located just nasal to theoptic disc. The zone of detachment inthis eye included the macula and asurrounding zone temporal to theoptic disc. (B) Seven days' detach-ment: 3 days' reattachment. The outersegment layer is virtually absent, al-though small stacks of discs can beidentified at the tips of some rod andcone inner segments. (C) Seven days'detachment: 5 days' reattachment.Most rod and cone inner segmentshave a short outer segment at thisstage. Vacuolation at the basal surfaceof the RPE shown here, and to a lesserextent in (D) and (E), is an artifact at-tributable to the perfusion fixation. (D)Seven days' detachment: 7 days' reat-tachment. In this parafoveal region,the ROSs and COSs have elongatedsomewhat, but the disc stacks remaindisorganized. The scalloped apical sur-face of the RPE {see E also) is charac-teristic of retinal regions that had beendetached. (E) Seven days' detachment:14 days' reattachment. After 2 weeks,ROSs and COSs remain substantiallyshorter than normal (A-E, X1000).

INVESTIGATIVE OPHTHALMOLOGY tjl VISUAL SCIENCE / August 1 989 Vol 30

q.4

Fig. 4. Seven days' detachment: 3 days' reattachment. At the shortest reattachment interval examined, many inner segments have no associated outer segment. Of those that do, the disc membranes are usually highly disorganized. A small minority of photoreceptors have well organized stacks of disc membrane (arrow) although the rudimentary outer segments are much shorter than normal. RPE phagosomes are rare at this stage (X4200).

Fig. 5. Seven days' detachment: 5 days' reattachment. At 5 days more inner segments have an associated outer segment. Large whorls of outer segment membrane appear in the subretinal space (arrow). A few aligned ROSS are apparent. The phagosome content of the RPE is very low. The swelling at the basal surface of the RPE (asterisks) is artifactual (X4200).

1716 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Augusr 1989 Vol. 30

Fig. 7. Seven days' detachment: 14 days' reattachment. After 2 weeks, ROSs and COSs have elongated somewhat, and their tips abut theapical RPE surface. Disc packets in the process of detaching from the outer segment tips are clearly visible (arrow) as are freshly shedphagosomes (arrowheads) (X4125).

12A through D are representative COSs selected fromthe retinas reattached for 3, 5, 7 and 14 days, respec-tively. Although the degree to which they regainedtheir normal morphology was highly variable, bothROSs and COSs tended to be more cylindrical at the7- and 14-day intervals. At the earliest timepoints, theCOSs consisted of only balloon-like extensions of thecilium (Fig. 12A); in other cases the outer segmentwas comprised of a short stack of closely apposed discmembranes (Fig. 12B). Stacks of outer segment discmembranes were frequently located at the lateral

margin of inner segments instead of being positionedat their distal tips. When this occurred, the connect-ing cilium was displaced laterally from its normallocation.

Recovery after Reattachment: The Photoreceptor-RPE Interface

At 3 days after reattachment, short microvillousprocesses were present on the apical RPE surface, butthey did not interdigitate with the rudimentary outer

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Guerin er ol 1717

Fig. 8. Frequency histo-gram of ROS and COSlength as a function of reat-tachment interval. (A)Reattached ROSs; (B) reat-tached COSs. ROS andCOS lengths were measuredin areas of reattachmentusing the protocol describedin the text. Lengths weremeasured in normal controleyes and at 3, 5, 7 and14 days after reattach-ment. Measurements weregrouped into 2 ̂ m bins andplotted as a percentage oftotal frequency. Inner seg-ments with no apparentouter segment were notcounted. Overall, bothROSs and COSs tended toelongate, and their distribu-tions widened considerablywith increasing time of reat-tachment. At the 3-daytimepoint, both ROSs andCOSs measured about 10%of their normal adultlengths. At the 2-week time-point mean length of reat-tached macular ROSs wasapproximately 40%, andreattached macular COSswere about 35% of normalmean length.

Reattached Rod Outer Segments 1001 Controls: • MaculaG Periphery

100-i

g§• 40-

Dav380-

60-

40-

20-

Day5so-

60-

20-

Dav7

40-

20

Dav 14

20-n = 49mcan = 19.2scm = 0.24

n= 110mean = 9.9sem = 0.23

n=130mean = 8.7scm = 0.24

n= 118mean = 6.4scm = 0.18

B o6 10 14 18 22 26 30 34 " " 3 ? , ,mean = 3.1

length (urn) scm = 0.17

Reattached Cone Outer Segments 1 Control: Macula

20-

Dav3

B 2 6 10 14 18 22 26 30 34 ^ 2 3

length (nm) scm = 0.14

segments (Fig. 4). Newly formed outer segments wereattached to their respective inner segments by a con-necting cilium (Fig. 12A). There was considerable la-mellar debris in the subretinal space, but no evidenceof phagosomes within the RPE cytoplasm. Pigmentgranules remained within the RPE cytoplasm. Nu-merous phagocytic cells filled with lamellar debris

were positioned close to the photoreceptor inner seg-ments.

Within the first week following reattachment, theRPE-photoreceptor interface underwent a significantreorganization (Figs. 4-6). Lamellar debris in the sub-retinal space decreased substantially, although therewas no evidence that cells of the RPE monolayer were

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Guerin er ol 1719

Fig. 11. Immunolocalization of opsin in regenerating ROSs. (A) Seven days' detachment: 14 days' reattachment. ROSs from reattachedregions of several retinas were uniformly labeled with Au particles, suggesting that opsin was probably transported and inserted into new discmembrane during the reattachment interval. <B) Higher power view of disc membranes shown in (A). Opsin was localized to the discmembranes of reattached ROSs using the immunogold, post-embedding technique described in the text (A, X 10,741, B, X25,187).

Persistent Outer Segment DefectsIn areas where most outer segments had regained

their typical cylindrical configurations, some photo-receptors showed persistent defects in outer segmentalignment and orientation, ciliary location, and indisc alignment (Fig. 14A, B). In cones particularly,large disorganized whorls of membrane located im-mediately above the inner segments were sometimesthe only indication of outer segment recovery.

DiscussionThe experimental detachment procedure used in

these experiments, direct microinjection of physio-logical saline into the subretinal space, produced arelatively atraumatic detachment of the macula andsurrounding retina. In acute detachments, separationof the two tissue layers resulted in a significant ex-pansion in the volume of the subretinal space and inthe formation of a dome-shaped bleb, without pro-ducing substantial morphological damage to eitherthe retina or RPE cells. The apical cell surface of theRPE cells changed abruptly after separation, but the

cells remained viable and adherent to Bruch's mem-brane. Experimental detachment, followed by a7-day waiting period, led to nearly complete degener-ation of photoreceptor outer segments, but left innersegments and the rest of the photoreceptor cells in-tact. Thus, at the time of reattachment in these ex-periments, photoreceptors had either no outer seg-ment at all or a rudimentary outer segment com-posed of a small amount of disorganized discmembrane. In addition, the specialized arrays of mi-crovillous processes that normally ensheath the outersegments were absent.

The earliest phase of morphological recovery thatwe observed was 3 days following reattachment (7days' detachment; 3 days' reattachment). At this stagethe photoreceptor cells were closely apposed to theundifFerentiated apical RPE surface. The subretinalspace contained a large amount of membranousdebris derived from the degenerated outer segments.Phagosomes were absent from the RPE cytoplasm.Outer segment regrowth was minimal but more conespossessed rudimentary outer segments than did rods.Electron microscopic autoradiograms strongly sug-

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Gu&m er al 1721

Fig. 12. Representative COSs at various reattachment intervals (A) seven days' detachment: 3 days' reattachment. Most COSs at 3 days consist of small stacks of apposed membranes that are connected to an adjacent cilium. (B) Seven days' detachment: 5 days' reattachment. Accumulations ofdisorganized membrane at the tips ofcone inner segments were relatively common in the early stages of regrowth. (C) Seven days' detachment: 7 days' reattachment. (D) Seven days' detachment: 14 days' reattachment. After 1-2 weeks, some COSs had a well organized disc stack. Processes from the RPE ensheathed the OS tips (A, X28.750, B, X30.187, C. X 15.675, D, X 13,650).

gested that the formation of these rudimentary outer segments occurred during the 3day reattachment in- terval. Regenerating rod outer segments, when pres- ent, tended to be more cylindrical than the cone outer segments, which often appeared as disorganized membranous stacks. These observations suggest that, although the initial regrowth of COS membrane may proceed more rapidly than in rods, cones do not reacquire their normal configurations as readily. The implication is that the mechanism by which rod and cone outer segments acquire and maintain their characteristic shapes is different. In light of the known differences between rod and cone outer seg-

ments in outer segment organization.l6," shape and extracellular milieu.'* such a difference is not sur- prising.

As the recovery process proceeded, the initial out- growth of outer segments coincided with a remodel- ing of the RPE's apical surface. At approximately I week after reattachment apical RPE microvilli were longer and more numerous and, for the first time, some processes interdigitated with the regenerating outer segments. Pigment granules were repositioned near the apical cell surface from their formerly ran- dom distribution within the cytoplasm, although they did not appear within the apical processes as is nor-

Fig. 13. RPE proliferation in the reattached fovea. Seven days' detachment: 30 days' reattachment. (A) Low-power light micrograph of the foveal pit. (B) Higher-power view of the region in (A), upper left, that is enclosed by brackets. An area of RPE proliferation immediately adjacent to the foveal pit. No COS regeneration has taken place in the region underlying the proliferated RPE cells. In addition. some thinning of the cone inner segments and nuclei appears to have occurred. (C) Higher-power view of the region in (A). upper right, enclosed by brackets. In the absence of RPE proliferation, some rudimentary COSs are identifiable: cone inner segments and nuclei are intact (A, x72. B, C. ~ 4 8 0 ) .

1722 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / August 1989 Vol. 30

Fi. 14. Persistent abnormalities in photoreceptor recovery 2 weeks after reattachment (7 days' detachment: 14 days' reattachment) (A) Abnormalities in the normal orientation of photoreceptor outer and inner segments is a common feature after reattachment. Cone inner segment (CIS), rod inner segment (RIS). (B) Some COSs fail to reacquire their normal cylindrical configurations and appear instead as disorganized whorls of membrane (A. X8225, B, X 13.750).

mally the case. In contrast to the earliest timepoints, membranous debris was no longer present in the sub- retinal space. having been phagocytosed by migrating phagocytes derived from two sources: RPE cells that detach from Bruch's membrane and blood-borne monocytes that invade the subretinal space from the choriocapillaris. At this stage, there was a noticeable increase in the number of rod inner segments that had associated outer segments. ROSS were usually longer than COSs, although most outer segments re- mained <50% of normal length. The outer segments in most, but not all, regenerating rod and cone outer segments displayed the prototypical parallel align- ment and cylindrical configuration.

In autoradiograms, the appearance of a discrete front of radiolabeled glycoprotein in regenerating ROSS, together with observations demonstrating that this front was displaced toward the tips at longer in- jection-fixation time intervals, indicated clearly that disc morphogenesis was proceeding after short-term reattachment. Similarly, the labeling pattern identi- fied in cone outer segments was identical to the dif-

fuse pattern of incorporation characteristic of normal cones,I9 suggesting that vectorial transport and incor- poration of nascent protein into COSs had not been permanently impaired by detachment.

Recovery at the photoreceptor RPE interface after 2 weeks reattachment was marked by continued elon- gation of the cylindrical outer segments and their in- terdigitation with the newly formed microvilli that emerged from the apical RPE surface. There was less distinction between ROSS and COSs in their rela- tionship to the apical processes than there was nor- mally. Abnormalities in the orientation of individual outer segments, the location of the connecting cilium and the organization of the disc stack persisted in many recovering cells. As the regenerating outer seg- ments elongated, phagosomes reappeared within the RPE cytoplasm and evidence of disc shedding from the outer segment tips was observed. These observa- tions, together with the autoradiographic and im- munocytochemical data, indicated clearly that after only 2 weeks' reattachment, the regenerating outer segments had reestablished the two primary processes

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Guerin er ol 1723

that account for outer segment renewal: biosynthesisof disc membrane at the bases and disposal of oldmembrane from the outer segment tips.20

There are parallels between the remodeling of thephotoreceptor-RPE interface that occurs after reat-tachment and the maturation of these two cell layersduring embryonic development. In the developingmammalian retina, the outer segments appear first astubular extensions of the cilia and then become dis-organized lamellar structures.21'22 Only subsequentlydo they appear as well aligned stacks of disc mem-branes. The RPE lacks the dense carpet of apical mi-crovilli at early developmental stages and only later,synchronous with the development of large numbersof balloon-like outer segments, do the apical pro-cesses increase in length and number and begin toensheath the developing photoreceptors.21 Similarly,our results show that in the detached retina the refor-mation of an aligned disc stack does not take placeuntil the photoreceptors have been reapposed to theRPE. In addition, the redifferentiation of the apicalsurface occurs only after outer segment regrowth isinitiated. The recovery process following reattach-ment, therefore, is aptly described as an example ofmutual induction. In that respect, it is similar to theprocess described in the developing mammalianretina.21"25

Most of the morphological abnormalities identifiedin the reattached cat retina4 were also observed in thereattached primate macula. In this study, where de-tachment duration was limited to 1 week, changesassociated with longer-term detachment, such as sub-retinal gliosis, were conspicuously absent. However,areas of RPE proliferation were identified in reat-tached eyes at every stage of recovery and, in onecase, directly within the fovea (Fig. 13). As in the cat,outer segment regeneration underlying such areaswas uniformly poor. For the most part, however, re-gions of proliferation in the macula were sparse.These results are consistent with the conclusion thatthe frequency of morphological abnormalities is in-versely related to detachment duration; and theyimply that reattachment per se may be sufficient toretard or even arrest such changes after they havebegun. There is now good evidence from studies ofhuman reattachment patients that an exponential re-lationship exists between detachment duration andthe recovery of acuity, with longer durations resultingin a progressive decline in final acuity.9 Taken to-gether, these studies provide a strong and compellingcase for minimizing the time between detachmentonset and reattachment of the retina.

The current results, along with results from earlierstudies of experimental retinal reattachment in the

owl monkey,12 rhesus monkey,3 and cat4 indicate un-equivocally that the photoreceptors and RPE retain asignificant capacity for recovery after short-term reti-nal detachment that includes regrowth of outer seg-ments and their interdigitation with newly formedapical processes from the RPE. In earlier studies3 itwas suggested that an initial, supranormal burst ofdisc morphogenesis occurred immediately after reti-nal reattachment, followed by elongation of outersegments to normal length after approximately 30days. The results from the current studies provide noindication of an initial burst. Rather, they suggestthat outer segment regrowth occurs steadily, but thatthe rate of disc morphogenesis is somewhat slowerthan normal at least in the first 2 weeks after reat-tachment. In normal rhesus monkey ROSs, the meanrate of disc displacement is reported to be approxi-mately 2.5 /am/day.19 Measurements of disc displace-ment obtained from electron microscope autoradio-grams of normal control ROSs in this study confirmthose estimates. At that rate regenerating ROSs couldpotentially reach normal length within 9-13 days inthe absence of any disc shedding. As judged by thescarcity of phagosomes in the RPE, normal discshedding does not appear to make a significant con-tribution to outer segment length at the earliest reat-tachment time points (3 and 5 days). Nevertheless,mean ROS length at both these timepoints (2.3 nmand 3.2 /mi, respectively) is substantially below thatwhich would be expected from assembly at the nor-mal rate (7.5 jum and 12.5 fim) (see Fig. 8). The pre-liminary autoradiographic data are also consistentwith this conclusion. Data obtained from a single an-imal injected with 3H-fucose indicate that the meandisplacement over a 4-day reattachment interval,from reattachment day 10 to day 14, was 6.0 nm or1.5 Mm/day (Fig. 10).

Extrapolation of the results gathered from this andprevious studies of experimentally detached retinasto those encountered in human patients should beapproached with caution. There is a high degree ofsimilarity between the macular detachments pro-duced in monkeys and the retinal detachments en-countered in human patients. It is highly likely thatthis similarity extends to the recovery process as well.However, there are also some notable differences be-tween experimental and clinical detachments thatmay lead to a certain degree of dissimilarity betweenthe morphological recovery observed and that whichoccurs clinically. Experimental detachments are pro-duced more rapidly (<1 min) than most detachmentsencountered in human patients where subretinalfluid tends to accumulate over a longer time, andoften leads to a progressively larger detachment. Sec-

1724 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / Augusr 1989 Vol. 30

ond, several different types of fluid (eg, BSS, dilutedNa hyaluronate, enzyme-treated vitreous, serum)have been used to produce experimental detach-ments310'26 or to influence subretinal fluid resorp-tion,27 whereas in human detachments subretinalfluid consists of incarcerated vitreous and serumcomponents. Third, detachments in human patientsoften involve large retinal tears or breaks, in con-junction with bleeding from damaged retinal vessels.Experimental detachments produced by microinjec-tion result in virtually no intraretinal bleeding andproduce an extremely small retinal hole which proba-bly seals within a few minutes. The process describedhere applies to recovery that ensues after a maculardetachment of approximately 1 week. Detachmentsof longer, or perhaps shorter, duration would be ex-pected to affect the temporal sequence of recovery aswell as the incidence of persistent abnormalities (seebelow). Finally, the variable of detachment height,the distance between the detached retina and apicalRPE surface, needs to be taken into consideration.Previous experimental studies suggest that higher ormore bullous detachments are correlated with moresevere degenerative changes in the photoreceptorlayer.4'28 The detachments produced in this studywere bullous initially, but over the course of the next7 days, detachment heights diminished rapidly to thepoint where they were not ophthalmoscopically vis-ible.

These factors lead us to conclude that the processdescribed here should probably be regarded as an ex-ample of near optimum recovery. The RPE-photo-receptor layers are shallowly detached for most of the7-day period; they are separated for a relatively shortinterval and in the absence of other, potentially con-founding, factors. Human retinal detachments thatoccur as a result of trauma, those which are relativelybullous and/or persist for longer than 1 week, or thosewhere there is other coexistent pathology would beexpected to have a higher frequency of morphologicalabnormalities in the macula that would, in turn, re-sult in a reduced capacity for visual recovery.

The RPE-photoreceptor interface in the primatemacula retains a remarkable capacity for morpholog-ical recovery following a retinal detachment of shortduration (^7 days). ROS and COS regrowth beginimmediately or very soon after reapposition of, thephotoreceptors to the apical RPE surface and within2 weeks after reattachment outer segments regain ap-proximately 40% of their adult length, reacquire theircylindrical configurations and restart the process ofdisc shedding. The process of outer segment regrowthis analogous to that which occurs during initial devel-opment where outer segment elongation is a function

not only of the rate of disc morphogenesis but also ofdisc disposal through periodic shedding.22 Evenunder optimal experimental conditions and in theabsence of underlying pathology, abnormalities inouter segment structure and other morphological de-fects persist and are probably implicated in the rela-tively small, but consistent diminution in visual ca-pacity associated with macular detachments of shortduration.

Key words: retinal reattachment, retinal degeneration, reti-nal pigmented epithelium, retina, photoreceptors

AcknowledgmentsWe are indebted to Coopervision, Inc. (Irvine, CA) for

their gift of the Ocutome microsurgical system. Addition-ally we would like to express our appreciation to GeofferyLewis for performing some of the surgical procedures andassisting in others; to Kenneth Linberg for critical readingof the manuscript and for helpful suggestions; and to Drs.Dean Bok, Brian Matsumoto and Terry Schuster for pro-viding opsin antisera.

References1. Machemer R: Experimental retinal detachment in the owl

monkey: IV. The reattached retina. Am J Ophthalmol66:1075, 1968.

2. Kroll AJ and Machemer R: Experimental retinal detachmentin the owl monkey: V. Electron microscopy of the reattachedretina. Am J Ophthalmol 67:117, 1969a.

3. Kroll AJ and Machemer R: Experimental retinal detachmentand reattachment in the rhesus monkey. Am J Ophthalmol68:58, 1969b.

4. Anderson DH, Guerin CJ, Erickson PA, Stern WH, and FisherSK: Morphological recovery in the reattached retina. InvestOphthalmol Vis Sci 27:168, 1986.

5. McPherson AR, O'Malley RE, Butner RW, and BeltangadySS: Visual acuity after surgery for retinal detachment withmacular involvement. Ann Ophthalmol 14:639, 1982.

6. Chisholm IA, McClure E, and Foulds WS: Functional recoveryof the retina after retinal detachment. Trans Ophthalmol SocUK 95:167, 1975.

7. Cleary PE and Leaver PK: Macular abnormalities in the reat-tached retina. Br J Ophthalmol 62:595, 1978.

8. Tani P, Robertson DM, and Langworthy A: Prognosis forcentral vision and anatomic reattachment in rhegmatogenousretinal detachment with macula detached. Am J Ophthalmol92:611, 1981.

9. Burton TC: Recovery of visual acuity after RD involving themacula. Trans Am Ophthalmol Soc 80:475, 1982.

10. Anderson DH, Stern WH, Fisher SK, Erickson PA, and Bor-gula GA: Retinal detachment in the cat: The pigment epithe-lium-photoreceptor interface. Invest Ophthalmol Vis Sci24:906, 1983.

11. Anderson DH, Williams DW, Neitz J, Fariss RN, and FlieslerSJ: Tunicamycin induced degeneration in cone photorecep-tors. Visual Neuroscience 1:153, 1988.

12. Young RW and Droz B: The renewal of protein in retinal rodsand cones. J Cell Biol 39:169, 1968.

13. Anderson DH, Fisher SK, and Breding DJ: A concentration of

No. 8 PHOTORECEPTOR RECOVERY AFTER RETINAL REATTACHMENT / Guerin er ol 1725

fucosylated glycoconjugates at the base of cone outer segments:Quantitative electron microscope autoradiography. Exp EyeRes 42:267, 1986.

14. Rogers AW: Techniques of Autoradiography. Amsterdam, El-sevier/North Holland Biomedical Press, 1979, pp. 409-410.

15. Erickson PA, Anderson DA, and Fisher SK: Use of uranylacetate en bloc to improve tissue preservation and labeling forpost-embedding immunoelectron microscopy. J Electron Mi-crosc Tech 5:303, 1987.

16. Cohen AI: The fine structure of the extra-foveal receptors ofthe rhesus monkey. Exp Eye Res 1:128, 1961.

17. Anderson DH, Fisher SK, and Steinberg RH: Mammaliancones: Disc shedding, phagocytosis and renewal. InvestOphthalmol Vis Sci 17:117, 1978.

18. Johnson LV, Hageman GS, and Blanks JC: Interphotoreceptormatrix domains ensheath vertebrate cone photoreceptor cells.Invest Ophthalmol Vis Sci 27:129, 1985.

19. Young RW: The renewal of rod and cone outer segments in therhesus monkey. J Cell Biol 49:303, 1971.

20. Young RW: The renewal of photoreceptor cell outer segments.J Cell Biol 33:61, 1967.

21. Yamada E and Ishikawa T: Some observations on the submi-

croscopic morphogenesis of the human retina. In The Struc-ture of the Eye II, Rohen JW, editor. Stuttgart, Schattauer,1965, pp. 5-J6.

22. La Vail MM and Hild W: Histotypic organization of the ratretina in vitro. Z Zellforsch 114:557, 1971.

23. Hollyfield JG and Witkovsky P: Pigmented retinal epitheliuminvolvement in photoreceptor development and function. JExpZool 183:357, 1974.

24. Hild W and Callas G: The behavior of retinal tissue in vitro,light and electron microscopic observations. Z Zellforsch 80:1,1967.

25. La Vail MM: Kinetics of rod outer segment renewal in thedeveloping mouse retina. J Cell Biol 58:650, 1973.

26. Machemer R and Norton EWD: Experimental retinal detach-ment in the owl monkey: Methods of production and clinicalpicture. Am J Ophthalmol 66:388, 1968.

27. Marmor MF, Negi A, and Maurice DM: Kinetics of macro-molecules injected into the subretinal space. Exp Eye Res40:687, 1985.

28. Machemer R: Experimental retinal detachment in the owlmonkey: Histology of retina and pigment epithelium. Am JOphthalmol 66:396, 1968.